Patozone

Computing Fundamentals

 

Computer = programmable machine that can automatically carry out a sequence of arithmetic or logical operations

- consists of a CPU, nonvolatile / volatile / secondary storage and peripheral devices

- Central Processing Unit (CPU) - the brains of the operation; carries out instructions in microprocessors

- Nonvolatile Primary Storage - long term memory, aka read-only memory (ROM), but can be re-written c EEPROM; is directly accessed by CPU and contains BIOS

-- volatile memory lost when comp turned off

- Instruction Set Architecture (ISA) = low-level code a computer natively understands

- clock speed = speed at which a processor cycles (not used to compare different chips by itself, must also take into acccount IPC)

- Instructions per Clock (IPC)  = number of instructions a processor can process per clock

- cores = the more cores the more programs you can run simultaneously; # of CPUs inside a single processor pkg

- Single Instruction Multiple Data (SIMD) extensions = allows processors to perform same operation on multiple data simultaneously

- cache = fast memory that libes onboard the CPU itself

 

Motherboard (logic board in Apple terms) is the Printed Circuit Board (PCB) that holds most of the computers key components and provides connectors for peripherals; like the CNS of the computer

 

- CPU connects to Memory via main system bus (aka local or frontside bus)

-- buses are just wires that transfer data

- peripheral devices connect to Bridges (which integrate data and allows hardware to be interchangeable) via input / output buses (peripheral bus types like Peripheral component interconnect (PCI) bus, PCI-express, Universal serial bus (USB), and firewire), bridges also connect to system buses

- traditionally has a northbridge and a southbridge

-- northbridge connects directly to CPU by front-side bus, providing SDRAM, graphics processor and an interface to the southbridge

--- northbridge has been gradually merged into CPU

-- southbridge also connects to northbridge by its own (slower) bus, and is in charge of slower functions, such as interfaces for hard drives, removable devices and other hardware, internet, add-ons (PCI) and BIOS/EFI

 

Random Access Memory (RAM) = a kind of computer data storage lost when comp turned off; can be magnetic (hard disk, zip disk, floppy), optical (CD/DVD), or solid state (ROM, flash memory)

- computers only able to address memory as large as 2^x bytes (x= max bit depth of processor ISA, or bit depth the operating system was written in)

- more ram is better performance, but OS has a limit

- Dynamic RAM = main memory of system, everything loaded to RAM to be executed

- Video RAM = memory used in video display

 

Parallel: multiple bits (8, 16, 32) sent at same time, requiring multiple wires, thicker cables and complex connectors

Serial = individual bits sent as stream along single wire, which is easier to wire, but slower

"Hot Swappable" = ability of a device to be either connected or disconnected w/o having to turn off the device

 

Hard drive = device that stores nonvolatile, random access digital data; used to be spinning magnetic disks, now solid state drives more common, cannot be reached by cpu

- Serial Advanced Technology Attachment (SATA) connects the hard drive to the motherboard in new computer or to peripheral device, which is done by Integrated Device Electronics (IDE) in older computers

- Storage capacity can be in gigs or terabytes; SI bytes are to power of ten, while binary bytes are to power of 2, explaining the difference of why a USB seemingly has less memory the first time you put it in the computer (is not memory taken up by the USB, just a difference in the units of measurement [computers are binary] consumers not so much)

- hard drives can be spinning disks or solid state (uses microchips and is faster, but lower capacity and high cost)

- Graphics Processing Unit (GPU) - displays images, runs video operations; found in separate video card c own ram, if shares c RAM the capability is reduced

 

Redundant Array of Independent Disks (RAID) is a combo of disk drive components that can be distributed and replicated across drives to inc storage function and reliability through redundancy

 

Mean Time Between Failure (MTBF) is the predicted elapsed time bwt inherent failures of a system during operation

 

Dual Inline Package (DIP) = packing hundreds of transistors into a single electronic package allowing exponential increases in speed

 

Moore's law = the complexity for minimum component costs has increased at a factor of 2 per year

 

Video Graphics Array (VGA), older version that is analog connector carrying only only video signal

 

Digital Video Interface (DVI) still used, but being replaced by HDMI; carries only video signal

 

High-Definition Multimedia Interface (HDMI) carries video and audio signals and is used c HDTV and Liquid Crystal Display (LCD) monitors

 

Display Port seen c high end LCD monitors, carries only digital video signals

 

Single Instruction, Multiple Data (SIMD) can do ultra-fast operations

 

Hardware = physical components on which software installed

 

Hardware interfaces:

- Integrated Device Electronics (IDE; aka ATA or ATAPI) is MC and not expensive parallel hard drive c internal connection built for single access, simple and fast, but is slow for multiple simultaneous use

- Small Computer Systems Interface (SCSI) has an internal or external connection and is parallel; it performs better than IDE but is more expensive

-Universal Serial Bus (USB) is hot-swappable and can provide limited power to devices, has speed upto 4.6 Gbps

- IEEE 1394 (aka FireWire) is hot swappable c speeds up to 3200 Mbps

- Serial ATA is newer, c internal or external hard drives, speed up to 3 Gbps

 

Peripheral Devices

Some input (keyboard, mouse), some output (monitor, printer); must be periodically replaced (3-5 yr life of computers generally)

Monitors have been evolving, from the cathode ray tube (CRT) which was replaced by the LCD and others

- Twisted Nematic (TN) - MC LCD type are fast and cheap but have poor color

- In-Plane Switching (IPS) - more expensive and not as quick as TN, but has better color

- Patterned Vertical Alignment (PVA) - best color and good viewing angles, pretty quick, but high-end

- Active Matrix Organic Light Emitting Diode (AMOLED) are LCDs c good color and vieweing angles, used on mobile devices, but the organic elements have a limited lifespan 2/2 degradation

 

Networking can use variety of equipment to connect 2+ computer

-- a network can be classified by: 1) Scale, 2) Connectivity, 3) Physical topology, 4) Architectural/functional

 

1) Scale Classification

Personal Area Network -- inter-connectivity of IT devices in close proximity of individual

Local Area Network (LAN)-- network in a short geographical area, such as in a building, an office building, or group of buildings

Home Area Network -- a LAN at home

Wide Area Network (WAN) -- computer network over a large geographic area, usually connecting numerous LANs

Metropolitan Area Network (MAN)-- spans entire metropolitan area or campus, made of LANs, but smaller than WAN

Global Area Network - interconnected network of networks across an unlimited geographical area

Virtual Private Network (VPN) -- a private network built over a public network infrastructure that uses security mechanisms such as encryption to allow secure access from a different location, using the public telecomminocations network

 

2) Connectivity -- can be typed by the type of connectivity (usually broadly as wired vs wireless)

-- Twisted Pair Cables -- wired network with copper wires twisted in pairs, usually c 4 pairs for computer network cables. 2 types of twisted pair cables exist: Unshielded Twisted Pair  (or UTP, used in phone cabling), and Shielded Twisted Pair (or STP, used in Ethernet cabling)

-- Coaxial cables -- wired network type with copper or aluminum wires that are insulated to lower interference and distortion, such as in TV cables; speed usually 200 to 500 megabits per second

-- Optical Fiber Cables -- pulsed light transmission across glass fibers, each about the size of a human hair, that are in protective layers, the internet as a whole consists primarily as optical fiber cables; speed is usually >100 gigabytes per second

 

Wireless networks have a wireless signal transmission, and have different standards used in each scale of network

-- Wireless Personal Area Network -- consists of radio and light based transmission methods, such as WiFi, Bluetooth, and infrared communication

-- Wireless LAN -- has radio waves using established WiFi communication standards, like IEEE 802.11b

-- Wireless MAN -- methods similar to WLAN, but uses WiMax 802.16 communication standards

-- Wide Area Network -- has WiFi access to Internet or other networks

 

Wireless Cellular Network Types

-- Global System for Mobile Communications (GSM) -- global standard for mobile device communication, identifies users by their Subscriber Identity Module (SIM) card and has simultaneous voice and data up to 11 Mbps

-- Code Division Multiple Access (CDMA) -- this is more restrictive, has no SIM card, and relies on server-side "whitelist", CDMA is slower than GSM, and it cannot be used for data and voice simultaneously

-- Long Term Evolution (LTE) -- wireless standard c faster upload and download speeds based on Multiple-Input and Multiple-Output (MIMO) and CDMA technologies

-- Personal Communications Service (PCS) -- digital wireless phone service c emphasis on digital, personal devices and mobility

-- Digital Advanced Mobile Phone Service (D-AMPS) -- digital version of original analog standard for cellular telephone service, made by adding Time Division Multiple Access (TDMA) to each of the 3 channels

 

3) Physical Topology Classification

Impacts performance, access methods, cost of cable, type of cable, there are physical and logical classifications for topology

 

Physical topology -- emphasis on "physical" layout of the network bwt nodes, meaning how they are actually connected

- Line - each computer connected to the next in a straight line, network has beginning and an end; no Hub reqd, least expensive type

- Bus -- common backbone cable connects to each device

- Ring -- can be physical or logical ring, like a linear bus, but last computer in network is connect to the first, easy setup, token based, but lacks fault tolerance and difficult to remove and add devices

- Star -- each computer connects to a central hub, cables form a star configuration from the hub

- Tree - comination of a linear bus, and multiple star bus

- MESH - all devices connected to every other node on network, not only sending their own signals, but pass on signals from other nodes; is reliable and fault tolerant, but the most expensive 2/2 extra cable, difficult to trouble shoot

--- Ethernet supports Linear, Star and Tree topologies

 

Logical Topology -- emphasis on pattern of data transfer bwt nodes, meaning how they appear connected

- - Bus -- Node broadcasts to all nodes in network and each node check to see if message is for them

- - Ring - only one node can transfer data at a time, avoiding collisions, uses token to delineate which node is able to transfer data

 

Wireles Topologies

- point-to-point -- send signal from transmitter directly to specific receiver

- line-of-sight -- clear uninterrupted view bwt transmitter and receiver

- scatter / reflective -- both c infrared, no line of sight, bounce off walls or ceiling, uses devices to intercept and redirect

- cellular - cell tower to cell tower broadcast

- radio broadcast == RF freq waves (narrowband and spread spectrum)

 

4) Architectural / Functional Classification

- Peer- to -peer -- distributed architecture, no heirarchy, task allocation distributed to all network devices, usually used for file sharing (Napster, BitTorrent)

- Client-server -- tiered systems c clients and servers, clients depend on server for services, data, and application on the network, clients may to some or no processing

 

Private Network -- can communicate c all other computers on the same network, but not directly accessible from outside the network; it is typically based on the same technology as a public network and is a way computers can connect to outside networks such as the internet; an example is your home network

 

Internet Protocols -- rules (syntax and grammar); the Protocol stack requires rules at multiple layers

- TCP/IP is a combined transport/network protocol suite used by the Internet

- the protocol stack for the internet involves applications (HTTP, SMTP), Transport (TCP), Network (IP), Link (Ethernet or WiFi) and phyical transfer of bits on the wire

- Email uses SMTP or POP, web browsing uses HTTP, File  transfer uses FTP, Streaming sound uses RTP, H.323, Instant messaging uses IRC, streaming video uses RTP and remote login TELNET or SSH

-- intranets are private networks that use the same protocols as the internet, but are not publicly available, and also use TCP/IP and the same Protocol Stack as publical networks (an "extranet" is an intranet accessible through the internet)

-- Web 3.0 is the continuous integration of the web into our lives, and is the executable part of the web

 

Hubs = common connection point for devices on a network, usually used to connect different segments of LANs

Gateways = network node through which all data packets and flow entering into a network goes

Switch = filters and forwards packets bwt dofferemt segments of a network; not all segments get data packets

Router = forwards data packets, connecting at least 2 networks; usually at the gateway, forwards based on header

Network Interface Controller (NIC) piece of hardware that connects computer to the network

 

Open Systems Interconnection (OSI) model is conceptual model that describes the communications functions of a telecommunication or computing system, and ensures interoperability of various systems with standardized protocols, and partitions a system into abstraction layers, and is heirachical

 

- Network Interface Card (NIC) is circuit card that controls all network communication

- Virtual Local Area Network (VLAN) -

- Dial-up modems (modulator-demodulator) uses telephone; symbol rate measured in baud

- Ethernet (IEEE 802.3) is MC networking connection and is a local area network (LAN), which plugs into most router / cable modems or spreads via broadband router for internet

- WiFi is MC wireless LAN connection; uses radio tech, router sends signal c limited range; there are 4 types:802.11a, 802.11b, 802.11g, and 802.11n and 802.11ac (the current standard, can deliver high bandwidth) - 802.11ad would be even faster (6.76 Gbps)

-Bandwidth - digital bandwidth not same as bandwidth of radio/TV broadcast, as digital bandwidth is the AMOUNT of data that can be reliably transmitted along conductive medium; the max bandwidth determined by the transport media used and communication protocol

-- expressed in bits / sec, and more measures the transmission's fidelity (how much data can be packed in) and not the speed of data transport

- Input / Output Handler gets data from ouside and stores in input file or can also take data from output table and send to an ouside system; communication modes are TCP/IP (direct real-time communications) or batch file; the I/O handler knows nothing about the content of the message, just how to communicate it and the format that it uses to process

- I/O Processor breaks down the message into component data and store in DB or gets data from db and build output message; the Message Segmentation Methods (Format Types) are delimited (HL7) or Fixed Length Fields; the I/O processor responsible for dealing c message content, but is not involved in communicating it

- Cellular networks are Wireless Wide Area Networks (WAN), used by cell phones, has base stationc c antennas to transmit radio signals, usually slower than wifi, can be interfered c by distance, physical barriers, interference, and amt of network traffic

- 3G - mobile device networking technology, up to speeds bwt dial-up and WiFi; locked down by carrier

-- Verizon and Sprint use CDMA2000 and AT&T and T-mobile use UMTS, which are incompatible

- 4G - mobile device networking technology, faster than 3G but slower than ethernet; locked down by carrier; should be up to 4 Gbit/s

-- Long-Term Evolution (LTE) up to 100 Mbits/s

-- WiMAX up to 128 Mbits/s, but short range than LTE

-- High Speed Packet Access + (HSPA+) up to 56 Mbit/s

 

Peripherals - devices that can connect to the computer; can be input (keyboard) or output (monitor) devices; 4 kinds of internal slots:

1) Peripheral Component Interconnect (PCI) - replaced ISA16, still being used

2) PCI Express (PCI-E) - come in 1, 4, 8, or 16- lane configurations, has replaced Accelerated Graphics Port (AGP) as expansion slot of choice for graphics cards

3) USB - has 6 standard sizes (Standard A/B, Mini A/B, Micro A/B); Mini B and Micro B are MC

4) Firewire - standard for high-end video and audio eqpt; can be 6-circuit alpha, 4-circuit alpha or beta

 

Software is a set of coded instructions that the hardware can access and run,

- operating system software (OS; ie MacOS  or Windows, Linux [c Fedora, Ubuntu, Red Hat]); primary interface bwt user and comp, manages hardware resources, and is platform for user programs, started up by BIOS, but is specific for certain processors

- Basic Input-Output System (BIOS) starts up the hardware and the OS

- application software (user programs like Word)

- utility software (disk defragmenters)

- firmware (traditionally unrewritable, closely related to hardware; ie BIOS) allows hardware to initiate itself and talk to OS; programmed directly to ROM

-- firmware can be updated via peripheral storage device or through network connection

-- Electrically Erasable Programmable Read-Only Memory (EEPROM) allows the BIOS to be updated through downloads without having to send your computer into the manufacturer; may make vulnerable to malware

- User applications - can be stored on the client computer or on a server

- Drivers are the connect bwt software and hardware

-- ie sound cards, cameras,

 

Malware can be delivered via trojans, viruses, worms, or backdoors that delivers spyware, botnets, rootkits, backdoors, fraudulent dialers, sabotage, or ransomware

 

Security

- physical security (physically accessing a computer)

-- if physical security breached by unauthorized person, must assume that a machine is compromised even if person didn't log in

- social security (tricking people, ie phishing)

- network security - info sent via connections; use WPA2 encryption, and data should never best sent over open WIFI connection bc all data accessible to ISP or wireless carriers

- software - ie pw, , make sure all drivers are up to date, need good antivirus, do not download suspicious files

- data security - are you backed up? uninteruptable power; can be backed up c remote service online

 

Databases

 

Database = Persistent collection of data in digital form, organized to model information of interest in a way that allows info to be easily extracted, modified and utilized

- Just rows and columns of cells with data; can be either a list of CDs or a medical record for exmaple

- DBMS is a special system software that manages databases

- excel uses a basic DBMS, Wiki uses Linux-Apache-mySQL-PHP (LAMP) using mySQL as DBMS

 

4 classes of options for persistent storage- data input, data updates, data retrieval, data deletion

- first standardized DBMS was CODASYL in 60's, in the navigational db era, c hierarchical and network models organized in tree-like structures c parents and childs

- Relational DBMS (RDBMS) in 70's c 2-D tables c keys, that eventually evolved into SQL; and avoids constant reqriting of links and pointers in navigational models as db content changes

- object-oriented data and indexing in the 1980s, ie Java and C++

- Joomla and Drupal use LAMP stack, which handle high flow data better than RDBMS, for which NoSQL was made

 

Criteria for Reliable Persistent Storage

- CRUD and ACID

- an entity is an object of the real world with independent existence

 

Create Read Update Delete (CRUD)

Create - insert new records into a database; can alter database mapping; ie new row inserted in db table

Read - retrieving and searching records;

Update - modify the records or how the records are related to one another

Delete - permanently remove record from DB; must make sure there is no "orphan data" left over

- "cascading delete" useful bc deletes all records of low level, thus reducing / eliminating orphan data

 

Atomicity Consistency Isolation Durability (ACID)

Atomicity - database modifications must be all or nothing, or it fails; a failure can be in the hardware, system, application, DB or power failures

Consistency - only valid data written to DB, which can be at the DB-level or the application (user) -level

- propagation constraints (such as cascade deletion and referential integrity) allow you to decide what constraints to place on a DBMS

- cascade deletion is to remove all of a datatype that generates invalid data entries

- referential integrity is having all links working

Isolation - ability of DB to isolate a transaction or insertion of a record from other transactions until its execution in complete

- restriction of access to data that is currently being edited ("read-write locks")

Durability - system can recover losses from failures

 

Limitations of data and db include:

- scalability - ability to do more work when demand is increasing; scaling horizontally often better than scaling vertically (failure of one computer does not result in system failure)

- redundancy - replicating db is expensive, should run daily routines to run backups and check db integrity

- schema rigidity - hard to remove or add data field after creation, NoSQL may be more flexible

- model complexity - ie 1-to-1, 1 to many, many to many

 

Null values are data that were not available at time of data collection and is still unavailable or hasn't been entered yet

- not the same as an empty file, which means that data for the file does not exist

- null values can generate a critical value, and may need to be entered

 

Selection bias either excludes data entirely or makes it less likely that certain data subsets or population members are selected for study or QI projects

- fields can change over time and either contains new data or it has entirely changed

 

Database Models

- every database defined by models, like a set of instructions

 

Flat Model

Single 2D table (ie Excel), which are simple and MC of DB models, though are not robust, can duplicate data, and are single--user only

- other ex: Comma-separated values (CSV) and Tab - Separated Values (TSV)

 

Hierarchical Model

Data in tree-structure, with "parents" and "children"

- good for sorted and nested data, simple, intuitive, not too much data redundancy

- inefficient at lots of DB operations, low-level, low productivity

- ie: XML DBs, IBM Info Manag Systems (IMS), Microsoft Windows registry, MUMPS

 

Massachusetts General Hospital  Utility Multi-Programming System (MUMPS or M)

- programming system used at the Veterans Affairs (VA) hospitals - Vistable A (VistA); a hierarchical DB

- very low level computing, thus high-performance, but can also be too basic and non-functional

- uses pointers c simple parent-child relationship

- can have an emulation layer that intercepts SQL commands and translates it

- has "global" variables that stores complex documents as single nodes (and single numbers stored similarly); very simple and accessible info, does as well as the new relational DBs in terms of efficiency

 

Markup Languages (_ML; ie HTML, XML)
Hierarchical, with tags to define elements

- easily readable for humans

 

Network Model

Used by most object-oriented DBs, navigational,

- similar to hierarchical but children can have multiple parents and parents can be at any level of tree, naturally models real world, intuitive, flexible, fast data-retrieval, but low level, slow DB loadings, poor productivity

- endorsed and standardized by CODASYL

- ie: DECVAX DBMS, RDM embedded, object-oriented DBs

 

Relational Model

All data stored in table; each table represents an entity, that corresponds to real stuff, data connected bwt tables by sharing what they have in common

- dominant DB model used today, links multiple 2D tables using key values, the basis of every EMR available today, (except Vistable A)

- types of tables in pathology LIS DB can be broken down to Primary data (pts, specimens, parts, blocks, lsides), dictionaries (submitting physician, pathologists / staff, part types, billing codes, report formats), and system and administrative (session maintenance, histo logs, printing batches, menu elements, audit trails)

- are a collection of entities of 1 specific entity type

- the rows are aka "tuples" or "records"

-- each table must have an attribute that uniquely id's each row

- columns are aka "fields" or "attributes"

-- values in each column must be of same data type, and must be within an "attribute domain"

- the order of rows and columns does not matter

- schema define tables, fields, views, indexes, relationships, procedures, functions queues, triggers and other objects and specifies what the db can do and how operates

- determined by set theory, freely available, SQL standardized, can use multiple programming languages

- but has high system reqs, computationally complex

- ie IBM DB2, Oracle, Microsoft SQL server

- each pt can have 0, 1, or many specimens, each specimen must belong to one and only 1 pt, each specimen must have at least 1 part, each part must belong to 1 and only 1 specimen

- specimen table has a bunch of pointers and little actual data, which point to pt and MD dictionary tables and gets info from parts table

 

Entity-Relationship Modeling (ERM) is a was of making flowcharts that illustrate entity relationships

- diagram known as Entity-Relationship Diagrams (ERD)

-- entities are the "nouns" and the relationships are the "verbs"

- the Chen convention is that entity sets are rectangles, relationship set are diamonds, and attributes are ovals

- attributes connected to entities or relationships by lines; if attribute underlined, it is a "key" value, and attributes can only belong to 1 relationship or entity

-- in Chen model, if entity participates in relationship, is linked by a line; a double line bwt entity set and relationship set means that all entities in set must participate in at least 1 relationship in the set; an arrow from the entity set to the relationship set means that each entity in the set must participate in 1 relationship at the most; and a thick line bwt the entity set and relationship means that each entity in the set must participate in 1 and only 1 relationship in the set

 

Key Values

- aka a "key", is an attribute or set of attributes that determines other attributes

- primary key - attributes or set of attributes DMBS uses to id each row in a table

-- chosen from pool of superkeys (keys that could id each row in a table, and is narrowed to candidate keys (superkeys that are not redundant)

- choice depends on kind of data the table has

- the MRN is usually the primary key of the pt id table

- secondary keys are used only for data retrieval, and can be thought of as good search terms (Last name, date of birth, first name)

- foreign keys are attributes in a table whose bvalues are the same as those of the primary key in another table or is null, and can link tables together

 

Data Repository

centralized repository of info about data such as meaning, relationships to other data, origin, usage and format

- ex: the form used to enter info into a data dictionary

 

Mapping is a way of organzing or systematizing information

 

Electronic Data Interchange (EDI)

Enables exchange of data bwt disparate systems (such as blood bank eqpt and computer system), different computer systems in same facility, and different computer systems in different facilities / institutions (hospital to hospital)

- using standardized EDI reduces costs by eliminating need for customized interfaces; and means that data can be exchanged bwt larger number of institutions

 

 

Avoiding Data Anomalies by using Normal Forms

First Normal Form (1NF)

- table has >=1 candidate key, 1 of which is the primary key; no repeating groups in a table, and each row can be retrieved by using the primary key

- several attributes, such as name and age in a small table with pt info may predict sex (ie sex dependent of name and age)

- may have partial dependency (dependency that is part of a primary key) or transitive dependency (dependency not part of a key)

 

Second Normal Form (2NF)

Table must be in 1NF and have no partial dependencies, but can have transitive dependencies

- partial dependencies must be split into n tables, where n = 1 + (# unique partial dependencies)

 

Third Normal Form (3NF) and Boyce-Codd Normal Forms (BCNF)

3NF is a table in 2NF, but without transitive dependencies

- BCNF states that there must be 1 and only 1 candidate key per table, and there is no data duplication except through foreign key values

 

Higher-order NF available, but not necessary, best to just have all attributes dependent on primary key, and independent of each other, and that no row caontains 2+ multivalued facts about an entity

- best to have database normalized

 

Structured Query Language (SQL)

- pronounced "sequel", is a language that performs CRUS on relational db, and is an ANSI standard

-- there are some modern "NoSQL" (Not only SQL)RDBMS's that convert SQL to syntax used by NoSQL db in order to augment the relational db

- SQL broken down into Data Definition Languages (DDL) and Data Manipulation Language (DML)

 

DDL is  way you control columns in SQL, made of:

CREATE - set up database / tables / index, defines columns, but leaves table empty

- ie CREATE DATABASE (first command in starting db), CREATE TABLE (creates table and allows column definition), CREATE INDEX can speed up searches

ALTER - adds, deletes or modifies column properties, but still need DML to input data

- ie ALTER ADD (adds columns to pre-existing table or adds constraints), ALTER DROP (deletes preexisting columns or column constraints), ALTER COLUMN (alters datatype of preexisting column)

DROP - permanently destroys a database/table/index and deletes data

 

DML inserts and helps search data, made of:

SELECT - ie SELECT * FROM patient gets content of table "patients", but can be further filtered

- "*" is a "wildcard" that selects for any value; also, "%" substitutes for zero or more characters, and [^charlist] substitutes from a single character not in the character list

UPDATE - updates table columns; if WHERE is not used, then updates all records in a table

INSERT - puts records in a table, typically supplying values for all columns

DELETE - deletes stuff, similar to UPDATE, if WHERE is not used all records in a table are deleted

JOINS - queries data for 2+ tables based on foreign key relationships; different kinds of joins: inner, left (outer), right (outer) and full join

 

Data control language includes GRANT and REVOKE

 

Transaction control language has COMMT (save), ROLLBACK (reset), SAVEPOINT (rollback point), and SET TRANSACTION (name on a transaction)

 

Rich Internet Applications (RIA) depend on a set of software with an OS, Webserver, DBMS and scripting language (the OS matters the least)

- MC is Linux Apache mySQL, PHP (LAMP)

 

Object-Oriented Model

Uses object structures in low- and high- OOP languages; based on network model usually, though can be relational; it is intuitive to programmers, can be very high-performance

- OOPs advantages are inheritance and polymorphism

- Object Database Management Group ODMG standards were never truly implemented, poor interoperability, poor user tools

- ie Objectivity DB, MUMPs implementation (?)

 

Document-Oriented Model

Higher level than object-oriented, most NoSQL are this, no need to declare apriori what type of data  is acceptable in a record

- intuitive, doesn't need SQL bc so simple

- but not standardized, not very high-perf

- ie IBM Lotus Notes, Couch DB, XML and Apache

 

XML

Are document-oriented by definition

- schemas, aka grammars, are definitions of tags, elements and attributes, and can be Document Type Definitions (DTD), XML Schema, or RELAX NG

- 2 types of XML databases:

1. XML-enabled DB: accepts XML as input and gives XML as output

2. native XML DB: XML documents are the fundamental unit of storage; and can use XPath, XSLT, or XQUERY to query an XML document

 

JavaScript Object Notation (JSON)

lightweight text-based human-readable data interchange syntax derived from JavaScript

- uses curly braces; JSON is the doc type that CouchDB uses

 

Graph Model

Type of NoSQL db model resembling a network odel, except all node-to-node links ("edges") have intrinsic propertoes

- faster for large datasets, share the performance benefits of network and OODBMs

 

Entity-Attribute-Value (EAV) Model

matrix model where every entity can have a lot of attributes, but each attribute is a small subset, relying on meta-data systems to build relationships

- usually intuitive to data sets in medicine, implemented on top of RDBMS (thus can use SQL)

- need to evaluate metadata

- ie Cerner Millenium, Google App Engine

 

Dimensional Model

Adaptation of the RDBMS consisting of fact tables surrounded and linked by dimension tables that have descriptive fields, resulting in multidimensional tables, used a lot in business intelligence and data mining

- but are computationally expensive and has bottlenecks in operation, non-intuitive

- widely used in data warehouses

- star schemas are dimensional models implemented on top of a relational db's; and snowflake schema has more complicated dimensions (fractal-like structure)

- ie SAP Crystal Reports

 

In data warehouses, data is extracted, cleaned, transformed and catalogued to be made readily available for data mining and decision support

 

Other Programming Concepts

 

Stored Procedures

allows for assoc for SQL-type language to work on tables in a flat-file or relational db from within the db itself

- esp useful in web-based and thick-client settings

 

Thick Client

Clients run a comprehensive set of applications

- has limited server processing

Positives: better use of network resources

Negatives: Higher costs, need to upgrade each workstation

 

Thin Client

Clients run basic applications

- most of the processing at server

Positives: Lower cost, cross-platform, easier updates

Negatives: need powerful servers, higher network demands

 

Triggers

storage procedures automatically invoked when 1+ of a preset list of triggering conditions are met

 

IT Frameworks

-- COBIT -- IT governancy -- Control Objectives for Information and Related Technologies, Current version = COBIT5

-- SOX 404 -- Law -- Sarbanes-Oxley Law, Section 404 = IT operational control processes and change management

-- ITIL -- IT management -- Information Technology Infrastructure Library -- service strategy / description / transition / operation and continual service improvement

 

IT Management = daily activities, decision making, control of IT resources and processes aligned to the business activities and goals, services and the operational excellence

- IT management is tactical, oversight, implementation

 

Change control = process that ensures that potential changes are recorded, evaluated, authorized, and monitored in a controlled and coordinated manner

- critical for IT management and project management, that applies for changes, revisions, alterations, additions, enhancements, and upgrades to aspects of hardware, software or operations

- documentation required by regulatory bodies like CAP (such as change description, change reason, persons responsible, change categorization, approval signature, risk mitigation assessment, eval and monitoring outcomes

 

Project - temporary focused effort that leads to an pperational entity lasting a limited duration

- usually has ad-hoc cross functional group of people c defined set of resources to undertake an interdependent sequence of activities for a goal for a particular stakeholder

- in IT projects, usually around software system implementations or upgrades

 

IT project governance is outside of individual projects, and is a framework for making effective project decision making

- made of 3 parts: 1) decision making structure, 2) People in the structure, 3) Information that informs the decision makers

-- Governance goals ensure that project has been properly conceived, is executed well, that the project conforms to the enterprise IT governance, delivers the expected value, that expenses are in line and risks are related to the project and that the scope and boundaries of the project are in the IT governance strategy

-- good governance shows clear ownership and accountability, transparency in well-defined communication channels, responsive decision and efficient project management, participatory and fair inclusive of stakeholder needs

 

IT Project Management - application of knowledge, skills, tools and techniques used in project activities to meet the project reqs, and includes planning, organizing, monitoring and controlling the project activities

- many domains are managed in IT projects, such as scope, integration, risk, procurement, people, communications, quality, funding and time

 

IT Project Life Cycle:

Initiation --> Planning and Design --> Executing --> Monitoring and Controlling --> Closing

- there is a loop bwt Planning and Design --> Executing --> Monitoring and Controlling that feeds off each other

-- Intiation defines the goals / objectives, the scope, funding, personnel, space / infracstructure, define the project plan, training reqs

-- Planning and design -- assess the work req'd to meet the reqs, determine amt of funding, create a proposed schedule (consider a PERT (project Eval and Review Technique) or CPM (critical path method) diagram, detail an activation plan

-- Execution -- initiate activation plan, formalize communications, manage project plan activities

-- Monitor and Control -- update project plans, escalate issues to project governance, measure quality of the work activity, communicate issue resolution, manage constraints and issues

-- Close out -- complete all project documentation, launch/implement/Go-live activities, communication plan close out, leadership project executive summary, project review and lessons learned event, celebrate success

 

Project Change Control -- must be incorportated to project management

- during projects, can be good or bad reasons for a project to change, that makes a good change control process critical

 

 

Laboratory Information System (LIS)

 

Hardware and software that allow e-data processing and management functions necessary to run the lab

- essentially a set of definitions and procedures

- Laboratory Information Management Systems (LIMS) used by research labs; but both process and manage data

- digital and tele-pathology systems can be used

-- digital pathology can be used in education, case reviews and QA; can be Whole slide imaged (WSI) or static (ie gross camera, scope cams) imaged

- telepathology involves common points of use (grossing, intraop dx, histo), equipment (camera/scanners, computers c software, comm devices (phones), and certain benefits (time) and challenges (cannot see specimen)

- EMRs (Epic) are starting to take over the rolls of LIS and APLIS (ie Cerner)

 

HC information standards are important to minimize number and levels of errors, ensure consistent and accurate test results, ensure usefulness of data extracted to improve processes and systems across the lab, allow for integration and sharing of data bwt applications and platforms, and ensure consistent nomenclature for individuals, events, objects, and more

 

Components

 

Made of LIS software, Database Management System (DBMS), an operating system (OS) and hardware

 

OS (set of programs monitoring hardware and software resources) and hardware (computers, network equipment and other devices [printers, scanners, etc]) are the IT components

 

DBMS is "middleware" bwt applications and database; it is software c set of programs key to labs workflow that control access, storage, management and data retrieval ,

- able to handle details of physical storage of data, maintain indices and keys, support data addition, updating and deleting, support changes in conceptual db design, client - server distribution, authorization / user privilege / security, data integrity, data recovery, data locking, support a query language

DBMS structure :

- hierarchical - arranges data in trees c parent-child relations, written in XML, used by IMB Info Mgt Systems (IMS), Microsoft Windows registry and Oracle Berkley DB

- relational - multiple 2-D spreadsheats, using SQL syntax, MC model in HC

- object-oriented (uses objects in programming languages like Java, that defines objects as collections of attributes, and can be created in Objectivity/DB and Mass Gen Hosp Utility Multi-Programming System (MUMPS) implementation)

- document oriented - doesn't define data types a record can have or req that each record have same amt of info, but can create new info w/o requiring defined location for data to reside, used by Lotus Notes, Apache and Raven DB

- creates and administeretcs databases (in Excel, mySQL, designed to support specific database models, and not ealiy transferred to different DBMS

- data storage must have "I/O" (on/off) rapid accessibility to data (all access to DB must go through DBMS)

- data warehouses extract data from multiple sources and stores in single database and stores data indefinitely, allowing an org to analyze and study data (only the DBMS knows the details of physical storage of data)

- databases are sets of data that are organized to model info

 

LIS software interacts c end user on various systems: tracks specimens, drives instrumentation, manage workflow, track results, QA, data xchange, database storage, operational metrics

- early LIS was made as custom systems developed as department-specific systems focused mainly on reporting, and began to be sold commercially in the early 1980s

-Stand-alone, where each dept has its own software for their own needs (ie Cerner, Orchard, Sunquest)

- Enterprise-wide system integrated and all LIS and EMR come from 1 vendor and maximizes interoperability, allowing comparability, though is difficult to meet a labs individual needs (ie Epic and McKesson)

- LIS can also improve QC, assurance, efficiency, pt safety (printing labels, automated specimen ordering) and costs

-- Redundant Arrays of Inexpensive Disks (RAIDs) are sometimes used for this

-- Storage area neworks (disks stored remotely and accessed by network) or network attached storage (data stored on computers attached to network) are commonly used

 

Architecture

 

Depends on types of devices used and how they interact, which may be mainframe (host-based) or client  server (MC)

- in mainframe data center, a central computer holds all data and software, and controls all actions; user access by "dumb" terminals

-- more centralized, secure (protected against fire, theft, power loss), easy to maintain, but less user interaction and customization; can also be at an outside site

-- can use terminal simulation software to connect via devices

- in client-server, desktop "clients" makes requests to "servers"; servers have different functions: database stored on database servers, there are fax and interface serves

-- all have a database server, but can have interface server and print servers, all of which connect to the client LIS softwre

-- has many points of failure, so one can go down without losing whole network (all others have ingle points of failure; lose one and lose em all)

-- thin client is a client-server variant where a desktop computer runs a simple (thin) program that connects to intermediary thin client server, allowing easy updates and compatibility across platforms

--- thin client can be bad if the thin client server malfunction (no will will be able to access system)

-- called thin bc has limited functionality, but runs easier

- cloud-based - LIS housed on 1+ remote servers maintained by staff; accessed by clients via network / internet; has good security but less user control (client servers and thin client have greater risks)

 

Middleware connects data from analyzers, possibly from different manufacturers with different software, to LIS

- Adds functionality to lab by adding processing power and conditional logic

- Allows multiple processes running on the same or different computers to interact

- Auto-validation software - uses sets of rules to see if lab result from analyzed can be released w/o human review

- Reflex testing software - monitors certain test results and orders additional testing based on results of previous test

- Refining test results coming off analyzer (Post-processing data, data/result characterization, result interpretation)

- middleware supplements LIS functionality, and can even become incorporated into LIS

Inter-facility networks

Integrated Delivery Network (IDN)

allows multifacility LIS capabilities

- may be easier if standardized, but not always possbl

- outreach testing can provide services to non-affiliated clients (like reference lab testing)

- pairing up labs on a single system is $$ effective

 

Web Portals

Physician Office Lab Link (POLL)

- lab portal is a front-end system for clients enabling data exchange

 

Application Service Provider (ASP) implementation involves business contract and IT architecture

- outsources LIS management, hosted at remote site

- Service Level Agreement (SLA) part of the deal, defines the business contract terms

 

LIS core elements

Dictionaries

Data tables that lists varied info about tests, and thus define what shows up when tests reported, aka LIS tables, determine structure, rules, formats, entity names and conventions for info processing and control data that can be entered into LIS

- ie test definition dictionary shows container types such as test name / code, lab dep, results ranges, units, including other types that are customized for a lab

- "person" and "user types" dictionaries store LIS users' security levels

-- "person" types can be specific ppl ordering tests

-- "users types" are pathologist, resident, techs, etc

- table definition is one of the most important factors in LIS, all dictionary changes must be tested before use

 

Interfaces

Exchange of usable data bwt software/hardware systems; and control things like admissions, discharges, electronic order entry and transfers

- application interfaces exchange data bwt lab and physician/reporting system

- application / instrument interfaces bwt instruments (ie chem/heme analyzers to printers/iphone5) and software

-- interacts with lots of peripheral devices

-- can be broadcast (directly from orderer to instr) or query type (from LIS to analyzer, analyzer puts specimen on and reads barcode) instr interfaces

-- can be download-only (direct transfer of pt id and test order from LIS to analyzer), upload only (analyzer directly uploads to LIS) or mixed

-- uni- (ie down- or upload only, unsolicited transmitted of observation from LIS to EMR) or bidirectional (MD to EMR, EMR to LIS, LIS bact to EMR and back to MD)

-- LIS instrument analyzer make $$ for vendors

- software must be set up in both analyzer and LIS

 

Health Information Technology for Economic and Clinical Health (HITECH) act - authorizes incentives and payments through Medicare and Medicaid to hospitals and providers

- dictates how EHRs can be used effectively and securely to achieve specified improvements in delivery of HC

- "Meaningful use" of EHRs intended to adapt and use EHRs in ways that improve safety, quality and efficiency

- Federal Coordinating Council for Comparative Effectiveness Research identifies standards of comparative effectiveness and steps needed to gather information, and sets priorities for evaluating research investments and funding (3 Stages: Stage1 - Data capture and sharing; Stage 2 - Advance clinical processes; Stage 3 - Improved outcomes)

- healthcare information standards necessary to: transmit info in EHR, share clinical data electronically to measure outcomes, and use e-records for reports, billing and payments; done by standardizing vocabulary, structure, content, messaging, and security

 

HC effectiveness Data and Information Set (HEDIS) - 81 measures across 5 domains of care that measures performance on important dimensions of HC and service

Physician Quality Reporting System (PQRS) - allows individuals and groups to see if they meet quality metrics

Ntl Commitee for Quality Assurance (NCQA) - accreditation program that drives quality improvement

 

Interface Engines

- point-to-point interface architecture, specified number of systems need to be interfaced (represented as eqn: n*(n-1) for the total number of interfaces); if a single interface stops working in pont to point model, all other connections stay functional

- interface engine architectures route messages bwt systems and LIS

- if interface engine stops working will lose all communication bwt all systems (thus are designed c lots of redundancy to minimize failure)

- typically more than one inferace engines bwt LIS and lab instrumentation

 

Health Level 7 (HL7) is most universal reference interface specification for exchange of health data, and HL7 messages are the lifeblood of healthcare, exchanges data bwt software from different vendors

- non-profit founded in 1987, trying to set standard for info representation in hospital LISs; named based on the 7th level of the OSI Model (it is not the 7th version, ie there was no HL1, HL2...)

- Syntax = information structure; Semantics = meaning / encoding information

- defined sequence of message segments with pre-defined data elements necessary, and not message content, so that both systems understand

- HL7 v3.0 includes an XML schema called Clinical Doncument Architecture (CDA) that defines a standard approach to encode a clinical document

- Whole Slide Imaging (WSI) strives for Digital Image Communication of Medicine (DICOM) standard

- although is the standard, several versions exist, and there is progress towards a next generation projects, like HL7 reference information model (RIM)

- HL7 messages have fields, segments and messages

- triggers are events that cause message to be sent (such as inpt admission or cancel discharge)

-- segments are parts of an HL7 message that has one or more fields (such as pt ID, Dr ID, pt location, etc)

-- order message

-- results messages have order, observation request and  observation value

- an example message may have a message header,  Event, Pt id, and pt visit

 

- every node in a network has assoc dependency of interface implementation and design

- HL7 message constructs carry task-specific info, such as Pt admin (admission, location, transfer, discharge), E Practice Management (documentation of practice standards), LIS, Pharmacy, Radiology, Billing, Epidemiology

- HL7 is an attachment that makes sure data gets to the right place (whereas DICOM is structural packet)

- HL7 data types include things about: Dx info, Event info, Insurance, Orders, Pt info, Visit info

- Single Source of Truth (SSOT) is 1 way data transfer

- HL7 logistic event message types include Admit, Transfer, Discharge, Registration, Transfers, Canceling moves, and Merging Pts

-- code A18 - Merging is a special case that may be done manually in some cases 2/2 higher risk

--- merge miasma: merging of different pt records and not knowing if they reallly belong to same person

--- overlay: records of 2 different pts combined into single record (creates risk of mistreatment)

 

eXtensible Markup Language (XML) standards and XML Schema Definition (XSD) document structuring convention are effective and modern, HL7 has lagged behind comparatively (maybe 2/2 large variety of medical lingo)

- HTML and XML are open-source languages that have increasingly been used for health interfaces

- XMLs constrained syntax has pre-existing referential document called the XSD, which allows for effective data exchange without initial validation (makes it not necessary( with its constraints on permissible data elements

- Web services distribution language (WSDL) allows a requesting site to query online info about available resources such as info exchange operations, data element types and data element sequences

- now thought that WSDL will replace HL7

 

Match and Tag Heuristics / Algorithms

Pt id of outbound message must match that found in the receiving system (prevent error / mismatch)

- must withhold any data exchange with even slightest possibility of mismatch

- this Match and Tag matches up demographics bwt source and target c multiple rules

 

Australia and New Zealand use their own version of HL7, the Australian Healthcare Messaging Laboratory (AHML), which is more constrictive than HL7, but works

- Match and Tag algorithms ensures pt data goes to the right charts

 

Patient-Center data model ensures correct patient id at every stage, which is hampered by inablility to tranfer pt data bwt institutions

- LIS is accession-centric system c results linked to specific specimen tested, while EMR patient-centrically organizes data, and communication vwt the 2 requires pt id system, like ADT system

- lots of effort needed to set up and maintain interfaces

 

Service-oriented Architecture (SOA)

Perform tasks indirectly via communication protocol (ie TCP/IP) connecting network of comuters

- one system provides same service to all applications, independent of underlying technology

- web services result of widespread use of HTTP and XML, allowing communication by providing message structure

 

Simple Object Access Protocols (SOAPs)

- new web services include range of protocols and markup languages that allow info exchange

- allow syntactic exchange and similar to HL7, allowing service oriented architecture

- XML schema has detailed data restrictions

- disallowance allows development of messages that not only the sequence and data type constrained, but also the content of the message restricted to particular types or lists of possible values

 

Regional Health Information Organization (RHIO) is a possible solution for multi-regional health delivery sites (access for pts, MDs, pharmDs, insurers, little clinics or hospitals)

- Enterprise Master Patient Index (EMPI) is a Master Patient Index (MPI) that links smaller orgs together at a single institution, usually used to merge pt data from multiple sources creating a master index of patients

- MPI is dependent on non-unique test data supplied by pts, and can cause up to 1% of pts to receive the wrong tx; thus should be obsolete

- a MultiEnterprise Master Person Index (MEMPI) does the same thing for multiple organizations in a RHIO

- military uses social security number as identifier

 

Worksheets

the "logical conveyor belt" of the lab; aka worklists and logs

- every test is assigned to at least 1 test definition dictionary, and worksheets define what test is performed on what machine, test orders are batched to a machine to run tests

- can provide place for manual data entry and allow results to be analyzed, to decide on further testing

- lab workflow determined by dictionary and workflow interactions, and can be manually entered

- on the AP side of things, logs analogous to CP LIS worksheets (ie the list printed in the morning telling histotecs what stains to do)

 

LIS functions for Administration

 

Billing - creates invoices, submit claims, track receivables for labs

- reduces staff time c automated billing code validation (ICD-10), supports rules-based / automated systems

- can be cloud based c built-in coding and conversion ability and accounts recveivalbe software

 

Outreach - can be add-on or stand-alone, lets clinicians interact from outside system

- also tracts couriers, specimens, customer relationship management (CRM), alaytic tools,

 

- also QI, Business analytics, Regulations and Inspections, Outcomes reports mandates

 

LIS System Operations

 

Implementation plan must be developed, with pre-installation tasks developed by implementation teams, which set realistic timelines

- must buy hardware and software, and create dictionaries, worksheets and interfaces

- Conductor User Accpetance Testing and validation, training and implementation also necessary

 

Change control - process in which changes are recorded, eval'd, authorized and monitored

- mandatory per CAP

- documentation items include: change description, reason for change, person(s) responsible, change category, degree of change, signature / approval for change, risk assessment, eval / monitoring changes

 

LIS security - administrator is responsible for overall system security

- control types of access to data input / accessibility

- must control confidentiality and privacy

 

Tech support with experts that can help users solve problems

- should plan, design and manage processes and procedures to resolve problems

 

Maintenance must be ongoing in hardware and software with written procedures and as part of staff routine, with occasional system upgrades requiring plans and verification procedures

 

Error handling stuff that happens c LIS failure and mistakes made by personnel

- must have system for documenting and addressing errors

 

System backup - aka business continuity planning; part of disaster planning c short- and long-term backup and downtime and restoration procedures

- 3 most important kinds of backups:

1) Patient data backup - can be done c disk arrays, CDs, RAID, tape or solid state storage, virtual severs, clustered or networked computers,  and can also be done with clouding or snapshot data backup

2) Master table backup

3) OS backup

 

Management Reporting - includes billing, change log, collection list and completion reports, inventory, modified result log, recurring order definition log, rejected orders log, result delivery log, reimbursement problem report, QC-Levy-Jennings report, turn-around time (TAT), user action logs, and utilization reports

 

Total LIS operation costs - difficult to justify costs

- can estimate using total cost of ownership model, which takes into account software / hardware costs including updates, indirect costs such as downtime and user training, long-term expenses like hardware replacement and future scalability

 

LIS validation - should consistently lead to expected results; usually very labor intensive process

- all blood bank software must be validated under the Safe Medical Devices Act of 1990

- 3 main phases: prospective / developmental, acceptance and retrospective validation

-- these can be further broken down into these phases: project management / planning, system reqs def and specification, design construction / coding, user acceptance, operation and support, retirement

- SOPs with test plans should be developed before machines are installed, and should be updated / tested regularly

- stress tests and disaster plans should be made

- validation after certain changes is mandatory

 

LIS regulatory compliance

Must follow HIPAA, CMS and CLIA regulations

- blood banking subject to FDA Safe Medical Devices Act of 1990

- Voluntary Accreditation by CAP, TJC, AABB, ISO and COLA

 

Health Insurance Portability and Accountability Act (1996)

- prevents unauthorized disclosure and modification of pt data, ultimately responsibility of Chief Information Officer (CIO) and security officer, but up to everyone in health care (HC)

- 3 basic groups of covered entities (CE): HC providers, HC plans and HC clearinghouses

- Business Associated (BA's) can enter into contracts c CE's, and are also responsible for privacy

- Electronic Data Exchange (EDI), such as payments / claims / eligibility / referrals etc has been standardized and streamlined

- Privacy Rule regulates how data shared; helped by HITECH act

- Security Rule is federal standard for transferring ePHI, includes security management, workforce security, information access management, and security awareness and training

Administrative safeguards - policies including risk analysis and management, sanction policy and info system activity review, assign security officer, training, procedures, contingency plans, information access management, password management, contracts

Physical safeguards - facility access, workstation use, device controls,

Technical safeguards - access and audit controls, integrity, person authentication, transmission security

 

 

Workflow

 

order --> processing --> analysis --> report

 

3 phases:

preanalytic (from test order to lab reception), analytic (sending specimen to right place to result entry), and postanalytic (result reporting)

 

Pre-analytic

Patient Identification

High risk situations in patient id: siblings / twins, sound alike names, common names that we don't recognize as common, unconscious pts

- up to 15% of pts records have duplicate info from a different pt that costs the medical system up to $2.5M per year

- Medicare / Medicaid id errors and fraud costs system billions per year

- types of hc fraud includes HC coverage (ie using family members id to get tx), fraud (ie stealing insurance), id theft (can be from someone close), and stolen names (no good prevention mechanisms in place?)

- Biometrics based on who someone is, not what they have or what they say, and is very accurate and being developed for use across HC systems

-- iris scanning has high specificity (>200 DoF inhuman iris folding pattern and is used in several locations in the US; done c IR scanning and can be done at a distance; but can be screwed up by glaucoma, pigment leaks, and some ppl dont like it

-- fingerprinting can be of several types and useful across time and devices; 3D-readers most reliable

-- palm vein tech developed by Fujitsu; veins in palm visualized by IR and algorithm converts to unique identifier; separate templates used in each client network so adaptation necessary to render an identifier from one health system to another; used at several places in the US; palm veins grow and thus need to be reimaged up to 5 yo

-- pts may be skeptical of biometrics (assoc of fingerprinting and police); may be disused (lack of enforcement / incentives, familiarity c pts)

 

Wrist band created early in hospital stay, and used before procedures or procuring specimens, and links the specimen to the pt

- aliquots taken in lab linked to specimen, and ultimately to the pt

- tracking improved c barcodes and RFID tags

 

Bar Coding

Assoc c inc efficiency, streamlined workflow, label accuracy, better turnaround time

- better pt safety overall

- labeling c bar code begins at specimen collection

- allows automation and robotics (fewer errors, fewer staff, auto-verification)

- simplifies specimen sharing for multiple tests

- may be able to read foreign barcodes

-- universal bar codes should be adopted

- however, LIS upgrades and complex technology req'd for bar codes, and time and effort req'd for implementation, such as installation, training and downtime procedures

- other problems assoc c barcodes are human error (majority of error), non-compliance, hardware error, bar code error, misreads (1/1B in industry, 1/100k in medicine), lack of interoperabilitty

 

RFID tracking in blood products from donor to pt

- base of electronic records is first created, including pts id, info about the specimen, and creation of test tubes / blood bags

- donor paperwork, test tubes and blood bags put into RFID-tagged containers, which can log temp, pressure, humidity during shipping

- pickup events logged c RFID reader

- can be used in rapid re-inventory at blood bank

- many pros/cons of active vs passive RFID tags

- the frequency of the tags also has several advantages/disadvantages

- may have problems c pt id security, incompatible RFID systems, tag durability in pathology setting, system obsolescence

 

Test selected and order created - paper form (c stat tests) or Computerized Physician Order Entry (CPOE) in EMR (which transfers the test to LIS where workload and accession number generated and specimen billed), order transferred to LIS and worklists generated

- can have future orders, standing orders, add-on orders, order cancellation

- AP reqs on paper and usually general order

 

Specimen collected - must correctly id pt and label specimen (can use scanner to help verify correct test); inpt vs outpt differences, stat vs routine

- AP-LIS less involved in collection methods than CP

 

Lab receives specimen - AP vs CP reception

- various ways to send to lab (courier, pneumatic tubes, dumbwaiters)

 

Analytic

Specimen/test distribution worksheets - lab aliquots vs AP side accessions and grosses, stains

- specimen tracked throughout lab, sent to appropriate instruments

 

Test performance and results entry - need QC

 

Test interpretation - high/low?

 

LIS rules and algorithms - autoverification, flagging,  middleware, reflex testing

 

Clinical Lab Testing

Processors -

Centrifuges -

Decapper -

Aliquoter -

Analyzer -

Recappers -

Asset tracking - barcodes, radiofrequence ID, specimen location and monitoring status (Red means has passed stat time limit), retrieval of stored specimens. data analysis

 

Accessioning - scan barcode label or manually accessioned (needs new label for the lab; typed and mouse menu selection)

 

Triaging - specimens sorted in bins according to priority after scanned (stat vs routine, )

 

Specimen processiong - may need to aliquot or centrifuge; middleware connects LIS to processors, aliquoter, decapper, or centrifuge

 

Automated testing - middleware (controls information from and to LIS and analyzers), specimen direction (line tells LIS specimen location), initial testing (analyzer performs tests in order received from LIS, then uploads results to LIS), review/fu (LIS reviews results and adds interpretation, can do calculations [ie INR], autovalidation, reflex testing), QA (qc tests for analyzers, calibration, maintenance, records stored in LIS)

- accessioning and triaging not automated...

 

Manual testing - counting cells, petri dish results

- tech logs onto LIS, scans specimen or types ID and finally puts in the test results

 

Specimen storage - line recaps tube and sends to storage, which is catalogued for easy access, and/or staff puts it in fridge and manually inputs location

 

Post-Analytic Phase

Result reporting - formatting depends on setting found in LIS; which is sent to EMR via interface

- format MUST be easy to understand (must periodically check to see that everything is kosher)

- interface must take into account different components of a result and information format

- clinician needs to be aware of critical results, addendums, etc

- info must be sent to billing system

 

Areas for improvement in LIS - cost, need for more eqpt, time in entering and transmitting data, limited data analysis, lack of integration in reporting, lack of communication bwt MD and tech

 

Anatomic Lab Testing

Cassette makers -

Cryostats -

Scopes -

Cameras -

Asset tracking

Cannot be fully automated, need some humans there

- barcodes (ie 1D such as code 128 or 2D such as DataMatrix, QR code, or MaxiCode) and labels important

- reading 2 identifiers does not help prevent misidentification (leading to incorrect tx, transfusions, procedures, dx, longer stays, $$, adverse outcomes and lawsuits)

- RadioFrequency IDentification (RFID) have near flawless read rates, and can be passive (draw power from the readers) or active (have battery, cost more), requiring a reader which is hand held, pad, or conveyor belt (tunnel or doorway) readers

- handwritten stuff is error prone

- locates specimens, analyzes data, tracking varies depending on phase of analysis (ie during RFS)

- can route excess material to research / biobanks, maximize workflow, identify bottlenecks

- mislabeling assoc c manual labeling, hand matching and task batching

- the CLSI AUTO12-A documents offers guides on how to make a good specimen label

 

Accessioning - manual entering of data to LIS, needs case details (difficult to automate)

- accessioner must create or open case in LIS, then enters details from paper form, can categorize from menu, labels and cassettes and slides produced

 

Triaging - check priority, sort

- ie RFS, biobanking, large tissues need longer

 

Grossing - software for grossing, templates, microphone (speech recognition or transcriptionists), scanners

 

RFS - speed important, may be faster to write by hand vs scanned

 

Histology- scan cassettes, embedding, slide label printing, microtomes, stainers, coverslipper

- pathologist reads slides

- can have storage info coded

 

Result reporting - synoptics, submission to LIS, which goes to EMR, alerting clinician that report ready, creating addendum

 

Areas for improvement - cost, additional eqpt, lack of asset tracking, time to enter and transmit data, limited automation, lack of notification (subjective results), lck of access to clinical hx, limited auditing

 

Specialized Laboratory Testing

 

Blood Banking - LIS has info on donations of products and transfusions

- slightly different from traditional LIS bc products are regulated as medical devices by FDA, thus it must meet strict regulations for testing, distribution and software development as well as following a barcode standard (ISBT 128, which prevents duplications)

 

Molecular Diagnostics - usually use LIMS instead of LIS; supports research nature, complex workflows and inventory management, variable informatics pipelines, storage data and security reqs molecular lab requires

-

 

Human Leukocyte Antigen (HLA) testing - used in transplants; needs to include large quantity of clinical data and be able to work with other systems

- complex labs: Killer Cell Immunoglobulin-Like Receptor (KIR) genotyping, Panel Reactive Antibodies (PRA), antibody id, virtual crossmatching

- United Network of Organ Sharing (UNOS)

 

Cytogenetics - analyzes human chromosomes, usually does FISH testing also

- should support complex testing, store medical / family history, organize images for analysis, allow comparison of family's test results, and deliver detailed / understandable reports

 

Microbiology - scopes, Identifier (ID/AST), Mass spec, PCR, blood cultures

- more complex testing phases than other parts of lab and must accomodate large range of specimen types

- results can be interpreted in free text or as coded datagrams

- must also report communicable diseases

 

Point of Care Testing (POCT) - aka bedside or near-pt testing, fast ToT for dx, testing performed on handheld devices close to pt and outside the central lab, rapid results

- requires oversight by lab medicine, and clinician must know how to know when to use PCT, use eqpt and interpret results

- ex: glucose, blood gases/ electrolytes, clot times, PT/INR, cardiac markers, HbA1c, microscopy, FOBT, UA, rapid strep and influenza A/B

- earliest form of testing (ancient Egypt, India, tasted pee for glucose, also bringing monks urine who would tell if foamy, red, brown etc); bedside glucose in 60's was all the rage, many new tests have become available

Waived testing: cleared by FDA for home use, so simple you cant go wrong, no harm to pt

- HIV, influenza, glucose

- coming out with new wearable to track health

Non-waived testing: high-complexity, microscopy

Challenges in POCT- cost, necessity, quality, performance, regulation, lab impact, system management

- must be able to correctly id pt

- is there a risk of the pt performing the test?

- should be uploaded to EMR; has unique id

-- flow of data: POCT device, docking station, central data management system (DMS), interface manager, LIS, hospital information system (HIS), to EMR

- may have to enter results manually ... or can go straight to EMR from device

- current best practice is device enterprise solution; allow multiple devies to attach

- central lab testing is more accurate and reliable and cheaper

Solicited Results - receiving system has open order to wait for results; exclusively in central lab; closed loop audit trail

- improves compliance for orders and billing; testing accountability

Unsolicited Results - receiving system does not have an order; orders auto-generated on appropriate pt encounted and result posted; mostly in POCT; no consistent way to match results c order, usually are verbal orders, can cause multiple results per one test (one to many)

- Admissions / Discharge / Transfer (ADT) system allow point-of-care coordinators to see pt data, and track location of POCT testing

- only 1/10 hospitals includes all POCT data in EMR

- POCT1-A, and now POCT2A developed by CLSI

-- standard includes a POCT device interface (connects devices and docking stations to DMS by Device Message Layer [DML] or Device Access Point [DAP]) and Information system interface (connects DMS to EMR), which are both HL7 standardized

 

Point of Care Informatics - providing POC info to right person, place, time, and context

 

Benefits of Wireless in POCT - "Real time" transfer of pt results, update pt id on device, push configuration updates on wireless to device, enforce competency assessment and operator lockout on device

- real time can be misnomer, wireless can be slow, device sync can be slow, providers dependent on fast posting of results, location map in POC can be inaccurate, ADT message can vary

 

Post-analytic

 

Report generation and distribution - format dictated by EMR

- must be in acceptable form per CLIA regs

- can produce interim reports, cumulative reports, discharge reports or requisition/order-based reports

- AP tests can be put in system of auto-faxed to md

- management reports

 

Report correction - addenda / amendments

Report archived and specimen stored

 

Clinical Decision Support Systems (CDSSs)

software that gives MDs additional knowledge and pt info that is algorithmically filtered and presented at appropriate times during health-care delivery

- ie computerized alerts

- must be based on vocab standards and structured reporting