Rural Health 2.0 and User-driven healthcare Shoubhik Bose Engineer and independent researcher

Health What

Health Why Staying informed. Medical education. Collaboration and practice. Managing a particular disease. Sharing data for research.

v/s Collaboration

Health How telemedicine, electronic medical records mHealth Connected Health, The use of the internet by patients themselves such as through o messageboards, o blogs.

Rural health 2.0 Technology for low-infrastructure and remote areas.

Blood test without bleeding ? TouchHB Low-cost. No bleeding. Trained personnel not needed. Invented by: Myshkin Ingawale

Anatomy of a social network

Tabula Rasa ( Facebook group )

User-driven healthcare : Our implementation

Mathabhanga district in West Bengal Findings from a healthcare questionnaire

Absolute survey respondents V/S Commonly known problems

User-driven healthcare - What

User-driven healthcare- Why Individual patients should not have to get themselves shunted between multiple health professionals with multiple waiting times. The patient’s information travels through a network of dedicated health professionals rather than the patient having to travel physically.

User-driven healthcare- How 1.Collect health information. 1.Team of health experts discuss 'privatey'. 1.Produce a solution.

Dissecting the system

Screen grab: INPUT

Screen grab: OUTPUT

User-driven healthcare - features Hassle-free login using Google account. Multi-interface input. SMS and voice service from low-end phones ( from rural- seekers ) Anonymised patient names. Secure discussion boards. Reports-hosting. Solution recommendation engine. Android Application.

The proposed flow Patient sends a voice/text message. Message is sent to the server and made available on the web page. Medics get notified by /phone.

Input TIER 1 PATIENT CARE-GIVER

Input - The proposed (alternate) flow Dial a number and narrate the problem over the phone. Audio Speech-to-text engine. Text ( after translation ) UDHC

Inputs - methods Scanned narratives Transliteration in local language

Input - Transliteration

Some of the Patient inputs we have recieved over /web so far.

The input - How do we anonymise? Scientific names used instead of real patient names. Mashed up with pincode for geolocation Blurring of names in reports manually using image editors before uploading. ABELMOSCHUS480***ESCULENTUS

The input - Anonymisation ( proposed ) 1.Image editors native to the web-based application to allow image manipulation after upload. 1."Identifier detection" algorithms to automatically remove patient identifiers. 1.Image tagging. ( Something similar to Facebook? )

Patient consent form ( English ) Patient Consent form I give my consent for this information about MYSELF/MY WARD/MY RELATIVE [indicate correct description] relating to my/his/her health to appear in the ‘User Driven Health Care’ UDHC ‘clinical problem solving’ forum and web based electronic health record. I understand the following: 1.UDHC forum is an initiative to promote transparency and ethical practices in health care while qualitatively studying its processes to assess the role of various (hitherto unidentified) factors that influence healthcare outcomes. 1.UDHC network aims to qualitatively study and in a controlled online-forum, support the phenomenon of sharing an individual's disease related information with other multiple stake holders in healthcare such as similar patients, related health and other professionals to determine if this shared learning activity is beneficial in the individual's healthcare outcome. 1.My information will be published without my real name attached and UDHC ‘clinical problem solving’ forum will make every attempt to ensure my anonymity addressing me solely by my anonymized user-name. 1.I understand, however, that complete anonymity cannot be guaranteed. It is possible that somebody somewhere - perhaps, for example, somebody who looked after me if I was in hospital or a relative - may identify me. 1.UDHC ‘clinical problem solving’ forum will not allow the Information to be used out of context. The text of the information may be edited for style, grammar, consistency, and length. 1.The Information may be published in the UDHC ‘clinical problem solving’ forum and associated journals on paper as well as in the internet, which would be distributed worldwide. 1.Information displayed in the UDHC ‘clinical problem solving’ forum is not supposed to replace advice from the primary care physician of the patient. Signed Accept (Patient) Witness

Patient consent form ( Hindi ) रोगी सहमति फार्म I मैं अपनी स्वीकृति स्वयं के बारे में / मेरे वार्ड / मेरे रिश्तेदार / [ सही विवरण पर संकेत दें ] से सम्बंधित मेरे / उनके ( महिला / पुरुष )/ के स्वास्थ्य सम्बंधित समस्त जानकारी जो की UDHC ' नैदानिक ​​समस्या का हल ” मंच पर जो कि वेब आधारित इलेक्ट्रॉनिक स्वास्थ्य के आधार पर रिकार्ड कि गयी है देने को तैयार हूँ | मैं समझता / समझती हूँ कि : UDHC netnet नेटवर्क का उद्देश्य गुणात्मक अध्ययन और नियंत्रित ऑनलाइन मंच है जो कि कई तरह के रोग संबंधित जानकारी कई अन्य स्वास्थ्य सेवा पेशेवरों के रूप में हितधारकों के साथ एक व्यक्ति की रोग से संबंधित जानकारी को बांटने की घटना का समर्थन करने के निर्धारित करती है कि इस गतिविधि सीखने साझा करनाव्यक्ति के स्वास्थ्य परिणाम में लाभकारी है. UDHC नेटवर्क का प्रारम्भिक उद्देश्य स्वास्थ्य देखभाल में पारदर्शिता और नैतिक को बढ़ावा देना है जबकि इसकी गुणात्मक प्रक्रियाओं कि भूमिका का अध्ययन अब तक विभिन्न ( अज्ञात कारकों ) के का स्वास्थ्य परिणामों पर प्रभाव का आकलन करने के लिए है | मेरी सूचना, मेरे असली नाम संलग्न किये बिना ही UDHC ' नैदानिक ​​समस्या के समाधान ' मंच में प्रकाशित किया जाएगा और मेरी गुमनामी सुनिश्चित करने के लिए मुझे एक बेनाम उपयोगकर्ता के रूप में संबोधित किया जाए | हालांकि, मैं समझता हूँ, मेरी पूर्ण पहचान या पूरा नाम न छापने की गारंटी नहीं हो सकती क्योंकि यह संभव है कि शायद कभी किसी को ; उदाहरण के लिए मेरे या मेरे किसी रिश्तेदार के इलाज के समय वे रिश्तेदार जिन्होंने हमारी देखभाल कि हो वे मेरी / हमारी पहचान करने में सफल हो जाएँ | UDHC ' नैदानिक ​​समस्या को सुलझाने ' मंच को अनुमति नहीं होगी कि वे संदर्भ से बाहर किसी भी सूचना का इस्तेमाल कर सके | सूचना कि शैली, व्याकरण, स्थिरता, लम्बाई आदि संपादन कि द्रष्टि से संशोधित किया जा सकता है | सूचना को UDHC ' नैदानिक ​​समस्या को हल ' मंच और साथ ही साथ संबद्ध पत्रिकाओं के रूप में इंटरनेट में प्रकाशित किया जा सकता है जो कि दुनिया भर में वितरित किया जाएगा |. UDHC ' नैदानिक ​​समस्या को हल करने के मंच में प्रदर्शित जानकारी मरीज ​​की प्राथमिक देखभाल चिकित्सक की सलाह को प्रतिस्थापित नहीं कर सकते हैं. हस्ताक्षर / सहमति / असहमति

Inputs - representation [ list ]

Inputs - representation [ graphical ]

Inputs - pitfalls and workarounds Rural population has less/no access to the Internet. Language barrier – very few might know English. Computer illiteracy. How to use website forms? Narrate as voice using mobile phones. Transliteration. (and even translation) Not a problem, mobile phones will do! Medical Kiosks administered by computer literate volunteers.

Hidden layer TIER 2 MODERATORS MEDICAL STUDENTS CARE- GIVERS

The hidden layer - discussion

Output TIER 3 CARE-GIVERS CARE-SEEKERS MODERATORS

Evidence-based learning

Recommendation engine Patient posts health issue. System "predicts" similarity with a previously solved health case. Health experts take assistance of previously solved case to provide solution. The new solution gets added to the "repository" of previously solved cases.

Data from BMJ

Custom Algorithm: Tag-ifying the health narratives

Heuristic approach for predicting similarity Tag- ify Determine similarity between 2 narratives. o choose those which have a high number of similar tags. o Threshold of the percentage of similarity. Test with 20%, 40%, 50%, 60%, 80% tags' similarity and analyse whether the algorithmic results match the logically expected results.

Artificial Intelligence Google Prediction algorithm Quoting the Google Prediction algorithm Given a new item, predict a numeric value for that item, based on similar valued examples in its training data. Given a new item, choose a category that describes it best, given a set of similar categorized items in its training data.

"Google Prediction" for health narratives Given a new health narrative, predict a previously solved case from the archive, based on similar valued examples in its training data. Given a new health narrative, choose a health record that describes it best, given a set of similar health records items in its training data.

Scores from our Mathabhanga chapter where the UDHC team is led by Dr.Biswas. August 2011 November health cases discussed over 20 to 30 % patients travelled to Bhopal after initial consultation

Crowdsourcing the UDHC platform Mr. Kar Who can contribute ? Medical students. Engineers. Designers. Doctors. Moderators. Translators. Content editors. Care-seekers... and the list goes one. Anyone can help make build this encyclopedia of health issues.

References Swan Health 2.0 model, Emerging Patient-Driven Health Care Models: An Examination of Health Social Networks, Consumer Personalized Medicine and Quantified Self-Tracking, Int. J. Environ. Res. Public Health 2009, 6, BMJ ( 2012 ), BMJ Case Reports - Article archive by date, Retrieved fromhttp://casereports.bmj.com/content/by/yearhttp://casereports.bmj.com/content/by/year Biosense Technologies (2012), Myshkin Ingawale, Retrieved from UDHC ( 2012 ) Retrieved from Evidence based learning, 5 steps to evidence-based learning, Anatomy of a social network, Dave Gray,

Thank you, Questions and comments ? You can also get in touch Blog: code-cooker.blogspot.com Twitter: sbose78 Facebook: shoubhik.bose UDHC website : care.udhc.co.in