m-Medicine Jayanta Mukhopadhyay Dept of Computer Science and

m-Medicine Jayanta Mukhopadhyay Dept. of Computer Science and Engg.

Collaborators. . • • A. K. Majumdar Suman Kundu Soubhik Paul Ashish Shubham

Desktop-based e-Healthcare solutions are not adequate due to the dynamic nature of hospital environments • Popularity of accessing medical record using handheld devices at the point of care. • Portable. • Mobile Access of Information.

In a Hospital Environment • • Doctors are always on the move. Regular & Exceptional Events. Continual follow-up process. Heterogeneous Medical Information.

Requirements • Anywhere, anytime availability of medical records • Accessibility of EMR through various equipments – Static (PC) – Mobile (Laptop, PDA, Cell Phone) • • Secure information access Reminder services for doctors and patients Automated calls to doctors during emergency Interactive discussion with other doctors and staff with patient’s data.

Some Applications • Medical references. • Read Journal & Clinical Trial. • Medical (Dose) Calculator. • Electronic medical record access. • Emergency Messaging.

Some applications on mobile phones • Mobile Camera phones in Digital Soft Tissue injuries • Tele-Consultation for Orthopedic patients using MMS/Mobile. • 3 G Mobile links for consultation between a moving ambulance and the hospital base station • Mobile phone ready EHR System

Applications under consideration • Integration with EHR systems. • Emergency messaging. • Online video and data conferencing.

Work Flow in Hospital Environment Patient Arrives Radiological or Pathological Test Registration treatment Telemedicine EMERGENCY Registration Treatment OPD/DOPD Discharge Patient Indoor Admission

Integration with mobile based access system v. Integration of Indoor Patient Management System v. PDA based Wireless Medical Information Access System • Wireless Medical Information Access Server. • Client Tools for PDA v. SMS based Emergency Messaging System

Telemedik 2005 • Manages Patients Electronic Medical Records in a backend RDBMS • Connects multiple hospitals in West Bengal and Tripura for medical information communication • Online Video and Data Conferencing

Extending Telemedik 2005 for indoor patient management v. Integration of Indoor Patient Management System v. PDA based Wireless Medical Information Access System • Wireless Medical Information Access Server. • Client Tools for PDA v. SMS based Emergency Messaging System

System Architecture Clients Layer Application Layer Database Layer Personal Computer Indoor Patient Management (IPM) system IPM database PDA Telemedik system Telemedik database Wireless Medical Information Access system WMIAS database Mobile Phone Emergency Messaging system EMS database

System Architecture

imedi. K- A web-based system • i. Medik supports all the features of existing Telemdi. K installations. • Separate module handling presentation of the content.

Limitation of handheld devices 1. Small screen area. n Hard to view complete pages. n Increases scrolling and searching complexity. n Increases cognitive load. 2. 3. 4. 5. 6. CPU Speed. Limited Memory. Display resolution. Input Device (Stylus). Bandwidth.

Solutions offered • Client Server based approach • Data filtering • Partitioned image display for large images • Buffer management

Routine Blood test and grouping form as displayed on PC

Routine Blood test and grouping Form as displayed on PDA High Scrolling Required

Scrolling Required to view Telemedik forms on PDA Designed for PC. Avg. Max. Min. Vertical Scrolling. 4 times 7 times 2 times Horizontal Scrolling. 3 times 4 times 2 times

X-RAY Data Display High Scrolling Required

Our Approach 1. Creation of Medical Knowledge Base. 2. Prioritization of keywords. • Term Frequency Calculation • Most Frequently Suggested Test 3. Text fragmentation algorithm. 4. Adaptive partitioning for display of large images. 5. Image Caching Mechanism.

WMIAS (Wireless Media Information Access Server ) 1. Patient data browsing. a. Text Data. Context Sensitive Priority Based Text Fragmentation Algorithm b. Image Data. Fixed Partition Image Navigation Algorithm 2. Prescribing drugs and advice. 3. Recording of Patients Routine Checkup.

MEDICAL KNOWLEDGE BASE • Medical terminology of diseases, symptoms, test reports. • Normal range of each test report. • Abbreviations and Acronyms of medical terminologies. • Critical key word (Ex High, Low, Abnormal etc. ). • Prioritization of Keywords.

Term Frequency Calculation • Search engine: Google • Input Document sets: searched with Key words (Disease Name, Synonym, Abbreviation, ) (Cardiology, Heart Disease, Cardiac Surgery, Cardiomyopathy, …. ). • Priority Value: The frequency of occurrence of a keyword on the document set.

Most Frequently Suggested Tests • Input: Data from database. • Priority Value: proportional to the percentage of patients of a particular disease type are asked to perform that particular test.

Context Sensitive Priority Based Text Fragmentation Algorithm Present data in order of there relevance for diagnosis and treatment. • Context is the patient disease type, signs and symptoms. • Relevance is assigned in terms of priority value. {Relevance of a test report gets changed with the change of context. } • A fragment is a set of fields, which are having very close relevance for diagnosis and treatment.

Context Sensitive Priority Based Text Fragmentation Abnormal Data Fragment 1 Fragment 2

Dynamic Partition Image Navigation

Visualization Applications for PDA. • Viewing, Zooming and marking of image. • Skin Patch Viewing Application. • ECG Viewing Application. • Display of graphs and charts.

Adaptive partitioning of large images Region of interest Image Viewer.

Image Viewer.

ECG Viewer. Skin Patch.

Caching Mechanism • Multiple copies of the latest requested image is cached in the PDA memory. • It is found that, for 30% of the requests, the image is retrieved from local cache. • Image have to be zoomed in and out to reach optimum resolution for display. • PDA supports single window display.

Performance with Caching Image access time with caching mechanism 5 4 3 Sec 2 1 0 10 20 30 40 50 Image Size (Kb) 60 70 Access Time 4 3 2 Sec 1 0 10 20 30 40 50 Image Size (Kb) It is found that average. image access time reduces to 24% with the use of caching scheme suggested. 60 70

i. Medi. K: for mobile devices • The System needs to fragment the data q Fragmentation of frame based display q Fragment of summary page q Fragmentation of reports 43

System Architecture Desktop Client Page Request Mobile Client Page Request Page Response Firewall Presentation Layer Business Logic Layer Page Response Web Proxy Layer • A secure four tier architecture - Web proxy layer, Presentation Layer, Business Logic layer and Database layer. • Protects not only the medical database but also the application components. Database Layer Refer: Security Analysis and Implementation of Web-based Tele-medicine Services with a Four-tier Architecture By Amiya K. Maji, Arpita Mukhoty, Arun K Majumdar, Jayanta Mukhopadhyay, Shamik Sural, Soubhik Paul, Bandana Majumdar At CPMP 2008, Finland 44

Fragmentation of frame based display • Desktop- multi-frame based interface • Small Display-Multi-frame based approach is not suitable • Each frame represents as a single page • To access data user needs to navigate from one page to another 45

Fragmentation of reports • Patients reports contain many fields and data. • System divides the reports in small fragments depending on two parameters ØNo. of characters per row ØNo. of rows ØUses HTTP header for device specific information. ØUse abbreviations guided by a dictionary. 47

Fragmentation- An example

Contd…. . Abbreviation for the word blood Blood Sugar Test 49

Prescription Writing Form

Multimedia data in PDA • • Viewing & Marking of image Profile Marking application ECG Viewer application Display of Graphs and Charts

Skin Patch Viewer

ECG Data Display

Graphs & Charts Family History Tree Growth Chart

SMS based Semiautomatic Emergency Messaging System

Problem Description • Emergency event create alert that should be delivered to appropriate person at appropriate time. • Efficient routing and monitoring of alerts are key to quality and cost effective healthcare service. • Communication of vital signs and crucial medical information in the alert message.

Key Issues. • According to the requirement of medical event various alert types are generated. • Alert parameters that qualifies the person to receive an alert. • Availability of medical staff and their reachable device. • Efficient routing , monitoring and logging of Message. • Automate message generation process. • Large area of network coverage.

SMS based Approach SMS based system. • Anywhere and anytime availability. • Classification Messages. • Creation of medical knowledge base to transmit selective medical record in the limited space of SMS. • Creation of appropriate medical groups for handling different emergency case.

System Architecture Message definition library Event Manager Emergency Message generator Scheduled Message generator Message Formatter i. Medik Database Message Communicator Reminder message generator User & task manager Regular message generator Message manager Message queue Mobile Devices

Message Definition Struct message{ id : unique identification number U : level of urgency. R: response needed or not? T: deadline. Bd : message body. } id : message_ type, message_generation_time, date and serial_no.

Message Definition Alert Message (ALR) Struct L{ ward_name; room_no; bed_no; } Reminder Message (REM) (id, Bd). Regular Message (REG) Referred Patient Message (REF) Acknowledgement (ACK)

Medical Knowledge Base Send Some critical medical record in the message. • Rule Base Development. • Select medical records based on patients signs, symptoms and disease type. R 1: if ( case emergency) then (emr respiratory rate, blood pressure, pulse rate) R 2: if ( case emergency burn) then (emr % of burn, location of burn) R 3: if ( type surgical ) then ( emr haemoglobin count )

Message Format Create Message. Abbreviation Dictionary. • Medical Terminologies (BP : Blood Pressure RBC Count, ESR etc. ) • English • SMS (You : U, To: 2, Urgent Ur etc. ) References : 1. http: //www. treasuremytext. com 2. http: //www. ejalgaon. com 3. http: //www. funsms. com

Message Format Continued. . Highlighting abnormal data. • Color Coding. • Case Convention. UPPER CASE: HIGH PRIORITY. Mi. Xe. D Ca. Se: Me. Di. Um Pr. Io. Ri. Ty. lowercase: low priority. Examples: ALR 12: 32|01: 02 //High priority alert Al. R 12: 32|01: 02 //Medium priority alert alr 12: 32|01: 02 //Low priority alert

User and Task Management Profile: ( uid, D, N, {Si} ) //N Phone Number, S Duty Schedule. Event: ( eid, D ) { Trauma patient management, Surgery, Stroke Myocardial Infraction, Pre-operative care, Post-operative care } Task: ( tid, D ) { Example: Draw Blood, Administer Oxygen, perform CT Scan, Immobilize patient etc. } Duty: ( tid, uid ) Schedule: ( eid, {tid}, {Ri} ) Profile Duty Task E R Diagram Schedule Event

Emergency patient management Different Emergencies. • Accidental Emergency • Myocardial Infarction • Stroke • Burn Injury • Pediatric Emergency • Gynecological Emergency • Etc……

Accidental Emergency Management Emergency Location Pre Hospital Care Ambulance Hospital Emergency Group Primary emergency group Secondary emergency group

Emergency Patient Management Groups Emergency Case Primary Group Secondary Group Trauma Specialist. Ventilation Group. Respiratory Group. Circulation Group. Emergency Group. Cardiologist. Respiratory Group. Circulation Group. Emergency Group. Neurologist. Respiratory Group. Circulation Group. Blood Bank. Radiology Group. Neurology Group. Myocardial Infarction. Cerebral Stroke. Radiology Group.

Scheduling of long-term follow up care Routine preoperativ e testing. Presurgical visit Move patient to OT Post anesthesia care unit. Anesthesia Surgery Reference http: //omni. ac. uk The UK’s gateway to high quality Internet resources in health and medicine. http: //omni. ac. uk/browse/mesh/D 011300. html

Example Alert Preoperational alerts. (αid 1, Patient X, “radiology exam”, Central Lab officer, Normal, NA, Two days before surgery ) Operation Alert. (αid 1, Patient X, “Prepare Patient”, Ward Manager, Urgent, t h) (αid 1, Patient X, “Anesthesia”, Anesthesiologist, Urgent, t+1 h) (αid 1, Patient X, “Surgery”, Surgeon, Urgent, t+1. 2 h) (αid 1, Patient X, “Post operative care”, Post operative care officer, t+3 h)

Implementation Mobile Phone GSM Modem Emergency Messaging System Clinical Database Base transceiver station

Example Emergency Message ALR 14: 12 Attend Cardiac Patient Pri = Hi B. P. Increased Loc = Male Ward Room-102, Bed-14 BP =180/140 Tmp = 98 Pulse = 95 Hemoglobin = 8. 3 Dead Line = 20 Minuts Emergency message for attending a patient admitted in hospital. ALR 20: 10 Attend a trauma pat in emerg. Pri = Hi. Unconscious. External Bleeding Dead Line = 10 M. Age = 30 Y Sex = M. RR = 24 BP = 100/190 Pulse = 120. Emergency Message for attending a patient of accidental emergency.

Emergency Messaging Service using i. Medik • Sends SMS to doctors’ cell phones to inform him/her about any emergency or patient referral. • Follows the same multi-tier architecture • EMS server resides outside the firewall intercepting incoming / outgoing messages

EMS Architecture Link

Smartphones/PDAs in Teleconsultation • CPUs running at 400 MHz, flash memory of 256 MB, We use Win Mobile 6. 0 based. Cost about Rs. 11, 000. • 3 G offers enhanced communication speed upto 7. 2 Mbps

Video Streaming using a PDA

Drawbacks / Challenges • No shared Image marking/Annotation for effective tele-consultation. • No use of Audio/Video Publishing from the Mobile Client to Server for conference / consultation. • Limited to WLAN environments, global 3 G, allow only HTTP protocol. • MMS allows point to point transfer for images/data only, no broadcast. No storage for future reference

Proposed system WLAN / 3 G RED 5 Flash Media Server Medical Database LAN G N /3 LA LAN W USB / Bluetooth

Proposed system • Tested using Windows Mobile 6 PDA. • Built using Java. FX and Java. ME technologies • Makes use of RTMP / RTMPT protocol for communication • Red 5 Flash server(open source) and My. SQL database

Key Features • Shared Image Drawing /Annotations Canvas

Key Features • Audio/Video Publishing from Mobile Client to the Flash Server

Conclusion • Increasing uses of hand held devices in accessing medical information. • Integration of EHR with hand-held devices is essential. • SMS infrastructure could be effectively used for coordinating activities. • Increasing use of higher services of data and multimedia communication in mobile devices.