400 likes | 693 Views
Appropriate Technologies for Improving Healthcare in Emerging Markets Bill Thies Microsoft Research India. Joint work with Michael Paik, Manish Bhardwaj, Emma Brunskill, Somani Patnaik, Indrani Medhi, Kentaro Toyama General Electric October 28, 2009. Microsoft Research India.
E N D
Appropriate Technologiesfor Improving Healthcare in Emerging MarketsBill ThiesMicrosoft Research India Joint work with Michael Paik, Manish Bhardwaj, Emma Brunskill,Somani Patnaik, Indrani Medhi, Kentaro Toyama General ElectricOctober 28, 2009
Microsoft Research India Established January, 2005 Seven research areas Algorithms Cryptography, Security & Applied Math Graphics and Visualization Mobility, Networks, and Systems Multilingual Systems Rigorous Software Engineering Technology for Emerging Markets Contributions to Microsoft: MultiPoint, Netra, Virtual India Currently ~55 full-time staff, growing Collaborations with government, academia, industry, and NGOs in India Microsoft Research India Sadashivnagar, Bangalore http://research.microsoft.com/india
“Technology for Emerging Markets” Understand potential technology users in developing communities Design and evaluate technology and systems that contribute to socio-economic development of poor communities worldwide Collaborate with development-focused organizations for sustained, scaled impact Research Group Goals Computer-skills camp in Nakalabande, Bangalore (MSR India, StreeJagrutiSamiti, St. Joseph’s College)
Aishwarya Lakshmi Ratan – International Development Public Administration and Jonathan Donner – Communications Nimmi Rangaswamy – Social Anthropology Indrani Medhi – Design Kentaro Toyama (Group Lead) Computer Science – David Hutchful – Human Computer Interaction Rikin Gandhi – Astrophysics Multidisciplinary Research Society Society Group Group Impact Impact Understanding Understanding Individual Individual Bill Thies - Computer science Saurabh Panjwani - Computer science Technology Technology Innovation Innovation
Research Sites - Our projects - Other projects studied
Microfinance & Technology IT and Microentrepreneurs Sample Projects Microfinance PC + mobile Qualitative studies Business analysis Research only Microenterprise PC + mobile Mixed-method study Research only Information ecology of very small businesses Potential of technology to support microfinance Simultaneous Shared Access Featherweight Multimedia Kelsa+ Information access PC Qualitative study Usage analysis Pilot Primary education PC HCI User studies Software SDK General education Electronics HCI User studies Prelim research Free access PCs for low-income office staff Multi-user systems for educational Paper and cheap electronics for low-cost multimedia Text-Free UI Digital Green Warana Unwired Info systems Mobile Intervention Rural kiosks Pilot AgricultureVideo Intervention Control trials Pilot User interfaces PC Design User studies Guidelines Substitution of mobile phones for rural PC kiosks Video and mediated instruction for agriculture extension Text-free user interfaces for non-literate users
MultiPoint: Multiple Mice per PC • Highlights: • Demonstrated equal learning benefits as single mouse • Released as SDK • In use in >200 schools in Thailand, Vietnam, India, Chile, … Before MultiPoint After MultiPoint
Digital Green:Participatory Agricultural Extension • Highlights: • Demonstrated 10x improvement in cost effectiveness over traditional extension practice • With funding from the Gates Foundation, spun off as independent NGO to scale up across India Record Farmer Best Practices Mediated Screenings in Village
Towards Programmable Microfluidics • Microfluidics is a powerfultool for biology, but ofteninaccessible to non-experts • We are developing programmingand design tools to lower theentry barrier to microfluidics: • Architecture: first manipulation of discretesamples using soft-lithography [LabChip’06] • Computer Aided Design: first tool thatroutes channels, generates GUI [ICCD’09] • Portability:first mapping of single ISAacross different chips [DNA’06, NatCo’07] • Optimization: first efficient algorithm forcomplex mixing on chip[DNA’06, NatCo’07] • Programming: first high-level programminglanguage for biology protocols [IWBDA’09]
A Language for Biology Protocols with Vaishnavi Ananthanarayanan Key milestones: - Prototype implementation in C: 40 protocols, 3800 lines of code - First demonstration of biologist executing a formal protocol (IISc) - Populating OpenWetWare wiki with standard, bug-free protocols - Interested users at UC Berkeley, U. Washington, Duke, IISc In biology publications, can we replace the textual description of the methods used with a computer program? Enable automation by mapping to microfluidic chips BioStreamLanguage Improve reproducibility by generating human-readable instructions
Actively infectious 850K/yr Tuberculosis Remains a Global Problem Global TB Statistics • $4B/yr. is spent on TB control • 14M patients worldwide • 9M new cases/yr. • India has highest burden • 3M existing cases • 300K deaths/yr. Prevalence by Region Focus on India New cases 1.9M/yr. Current reach of care providers 450K/yr
Challenge: Medication Adherence • Tuberculosis patients mustadhere to a strict drug regimen • 4 drugs, 3 days / week, for 6 months • Consequences of missed doses • Not cured • Develop drug resistance • Medication adherence is a primary barrier to cures • Worldwide cure rate and W.H.O. target: ~85% • Barriers to adherence: • Side effects - Lack of education - Travel • Stigma - Expense of medicines - Forget / too busy Single day’s dose of TB medications Courtesy PIH Courtesy PIH
Directly Observed Therapy (DOT) Relies on providers to observe each dose Government clinics, hospitals Private providers: NGOs, community workers, traditional healers, business owners, etc. Protocol Patient-wise boxes (PWB) to provider Thrice weekly doses; patient travels to provider Provider must fetch patients who miss doses Supervisor visits quarterly to check progress One supervisor for 500K population (average load: 500 pts!) Providers get Rs. 250 per “successful outcome” Cured, or treatment completed but inconclusive Patient-wise boxes
Revised National Tuberculosis Control Program Implements Directly Observed Therapy (DOT), from 1997 Case detection: 61% Treatment success: 86% Since 1997: 6 million patients treated; >1 million lives saved Big program with tremendous impact Recent challenge: expanding into poor, remote communities Difficult to supervise and ensure quality Case detection reversed from 67% to 61% 25% of patients are retreated; 17% have MDR-TB 110K new cases of MDR-TB annually Challenge: Ensuring high treatment quality while scaling
Current Cornerstone: TB Treatment cards • Drawbacks • Hard to verify if visits happened • Hard to quickly interpret • Hard to aggregate • Treatment programs operate in the dark • Are drugs reaching patients? • Are patients taking medication? • Are patients getting better? Our Mission: Track Interaction, Adherence, Health
The uBox: A Smart PillboxDeveloped by Abiogenix, MIT, and Innovators In Health • The uBox monitors • Delivery, by logging patient/worker visits • Adherence, by logging pills dispensed • uBox impact • Worker supervision and incentives • Timely and targeted intervention • Lowers adherence burden uKey (one per patient,one per worker) uBox(one per patient) Patients Workers Clinic
A Biometric Terminal for TB Clinics • For verifying that patientand health worker interacted • Consists of: • Low-cost netbook (Rs. 12,000) • Fingerprint reader • Low-cost cell phone for data upload • Usage model: • Patient scans fingerprint upon each visit to the clinic • At the end of the day, visit logs uploaded over SMS • Data visualized by supervisors at central offices • Benefits: • Immediate response to missed doses • (As per uBox): incentives for workers, accountability to donors • Estimated cost: < $2 / patient
The uPhone: Monitoring Patient Health Worker relays vital patient health indicators using cell phone Nurseanalyzes data,identifies problems Physician sends advice to patients, schedules field visits Patient lives in a remote area
Is Technology Really the Answer? • Often ignores systemic and societal issues • But, delivery is overwhelmingly about diligence • Today: 2.4M doses/day, 187 countries, 85% reliability • Need: 7M doses/day, 100% reliability • FedEx: 7.5M shipments/day, 220 countries, 97.7% reliability • Our goal is to reduce the burden of diligence • Change the culture: 85% is not enough • Need to respond to every failed transaction • Identify superstar workers early and replicate techniques
Iterative Design: UBoxBihar, Jan. 2008 • Class proficient in less than 3 hours • Incorporated feedback into 9th design revision
Iterative Design: Fingerprint Terminalwith Operation Asha in Delhi, October 2009 • 4 days, ~30 patients served • Overwhelmingly positive response • Refinements: • Don’t use thumb print • Add incentives for providers,who sometimes relied onintermediaries to deliver drugs • Minor interface improvements • Excitement from partner, planning controlled trial
Iterative Design: UPhoneBihar, Jan. 2008 • uPhone more challenging • Despite intensive training, many errors on menu-based interface
Data Collection on Mobile Phones OpenROSA FrontlineSMS Forms [Banks] Nokia Data Gathering [Nokia] RapidSMS[UNICEF] MobileResearcher[Populi.net] Cell-Life in South Africa [Fynn] JivaTeleDoc in India [UN Publications] Pesinet in Mali [Balancing Act News] Malaria monitoring in Kenya [Nokia News] Voxiva Cell-PREVEN in Peru [Curioso et. al]
Data Collection on PDAs SATELLIFE EpiHandy EpiSurveyor[Datadyne] Infant health in Tanzania [Shrima et al.] e-IMCI in Tanzania [DeRenzi et al.] Respiratory health in Kenya [Diero et al.] Tobaccosurvey in India [Gupta] Ca:sh in India [Anantramanan et al.] Data Collection on Mobile Phones OpenROSA FrontlineSMS Forms [Banks] Nokia Data Gathering [Nokia] RapidSMS[UNICEF] MobileResearcher[Populi.net] Cell-Life in South Africa [Fynn] JivaTeleDoc in India [UN Publications] Pesinet in Mali [Balancing Act News] Malaria monitoring in Kenya [Nokia News] Voxiva Cell-PREVEN in Peru [Curioso et. al] Malaria monitoring in Gambia [Forster et al.] Clinical study in Gabon [Missinou et al.] Tuberculosis records in Peru [Blaya et al.] Sexual surveys in Peru [Bernabe-Ortiz et al.]
Data Collection on PDAs SATELLIFE EpiHandy EpiSurveyor[Datadyne] Infant health in Tanzania [Shrima et al.] e-IMCI in Tanzania [DeRenzi et al.] Respiratory health in Kenya [Diero et al.] Tobaccosurvey in India [Gupta] Ca:sh in India [Anantramanan et al.] Data Collection on Mobile Phones OpenROSA FrontlineSMS Forms [Banks] Nokia Data Gathering [Nokia] RapidSMS[UNICEF] MobileResearcher[Populi.net] Cell-Life in South Africa [Fynn] JivaTeleDoc in India [UN Publications] Pesinet in Mali [Balancing Act News] Malaria monitoring in Kenya [Nokia News] Voxiva Cell-PREVEN in Peru [Curioso et. al] Published Error Rates Malaria monitoring in Gambia [Forster et al.] Clinical study in Gabon [Missinou et al.] Tuberculosis records in Peru [Blaya et al.] Sexual surveys in Peru [Bernabe-Ortiz et al.]
Data Collection on PDAs SATELLIFE EpiHandy EpiSurveyor[Datadyne] Infant health in Tanzania [Shrima et al.] e-IMCI in Tanzania [DeRenzi et al.] Respiratory health in Kenya [Diero et al.] Tobaccosurvey in India [Gupta] Ca:sh in India [Anantramanan et al.] Data Collection on Mobile Phones OpenROSA FrontlineSMS Forms [Banks] Nokia Data Gathering [Nokia] RapidSMS[UNICEF] MobileResearcher[Populi.net] Cell-Life in South Africa [Fynn] JivaTeleDoc in India [UN Publications] Pesinet in Mali [Balancing Act News] Malaria monitoring in Kenya [Nokia News] Voxiva Cell-PREVEN in Peru [Curioso et. al] Published Error Rates Published Error Rates Malaria monitoring in Gambia [Forster et al.] Clinical study in Gabon [Missinou et al.] Tuberculosis records in Peru [Blaya et al.] Sexual surveys in Peru [Bernabe-Ortiz et al.] None? CAM in India [Parikh et al.]
Usability Barriers(Indrani Medhi) • Conducted ethnographic observations of 125 people on traditional text-based interfaces • Navigation difficulties: • Navigating hierarchical structures • Mapping soft-keys • Input difficulties: • Using scroll bars • Using checkboxes • Constructing SMS and USSD syntaxes • Language difficulties: • Specialized terms (e.g., transaction, jaundice)do not translate to local language In Ph Ke RSA
Design RecommendationsCase 1: Text-Based UI • Provide local language support (in both text and audio) • Minimize hierarchical structures • Avoid requiring non-numeric text • Avoid menus that require scrolling • Minimize soft-key mappings
Design Space flexible Free-form speech Live Operator Structured speech Input method Spoken Dialog Typing Graphical UI Text-Based Forms, SMS, etc. IVR Interactive Voice Response flexible inflexible Text Audio Graphics [+ Audio] Output method
Focus 1: Text vs. Spoken Dialog, Graphical UI Task: transfer money to a peer Participants: 58 non-literates (up to 6th standard), Bangalore Rich multimedia UI (without text) • Conclusions: • Non-text designs are strongly preferred over text-based designs • While task-completion rates are better for rich multimedia UI, speed is faster and less assistance is required on spoken-dialog system
Design RecommendationsCase 2: Rich Client UI • Recommendation: graphical UI with spoken input?
Focus 2: Text vs. Live Operator Task: report patient health symptoms Participants: 13 literate health workers and hospital staff, Gujarat Append to current SMS:11. Patient’s Cough:No Cough - Press 1Rare Cough - Press 2Mild Cough - Press 3 Heavy Cough - Press 4Severe Cough - Press 5 (with blood)— printed cue card— • Conclusions: • Live operator interface is only one with sufficient accuracy for health data • This model is also simple to adopt and cost-effective in India (call centers cheap) • Results caused partner to switch upcoming TB program from text to operator
Focus 2: Text vs. Live Operator Task: report patient health symptoms Participants: 13 literate health workers and hospital staff, Gujarat Health workers 7.6% 6.1% 3.2% Hospitalstaff 1.5% 1.3% 0% • Conclusions: • Live operator interface is only one with sufficient accuracy for health data • This model is also simple to adopt and cost-effective in India (call centers cheap) • Results caused partner to switch upcoming TB program from text to operator
Sources of Error Usability Barriers - small keys - correcting mistakes - decimal point - scrolling / selection - SMS encoding Multiple Choice (SMS) Numeric Multiple Choice (Forms)
Sources of Error Detectable Errors Usability Barriers - small keys - correcting mistakes - decimal point - scrolling / selection - SMS encoding Multiple Choice (SMS) Numeric Multiple Choice (Forms)
Design RecommendationsCase 3: Reporting Short Data • Recommendation (in India): use a live operator • Our proposition:Operators are under-utilized for mobile data collection • Benefits: • Lowest error rate • Less education and training needed • Most flexible interface • Surprisingly cost effective! • Challenges: • Servicing multiple callers
Overcoming Usability Barriers Place call Send SMS File Transfer via BlueTooth
Peer-to-Peer Media Sharing (Thomas Smyth) • Ethnography of Bluetooth video sharing in urban Bangalore Amongst non-computer owners (middle and middle-lower class) Over 25 in-person interviews, 100 phone interviews, 10 weeks • Rich ecosystem of downloading / disseminating videos • Despite steep barriers in usability, cost, legality, fear of viruses, privacy, etc., the drive for entertainment prevails. “Where there’s a will, there’s a way!” Mobile video sharing network
Conclusions Biometrics Smart pillbox Phone • We believe lightweight electronic devices have great promise in bringing healthcare to the underserved • In case of TB,monitoring deliverycould be key partof the solution • However, it is important to consider usability as first-class design constraint, as devices useless otherwise • Our status and next steps: • Refining biometrics prototype for controlled trial in urban Delhi • Working with partners to establish treatment program for trial of uBox, uPhone in rural Bihar