Our goal is to enable high-quality, sustainable medical data acquisition in settings where electronic medical record systems are not feasible and clinical insight is needed most.
Our pilot study on the Thai-Burma border delivered data-driven insights to better understand maternal health outcomes among migrant mothers at the Mae Tao Clinic.
Although free and open source medical record software exists, the vast majority of clinics and hospitals in low-resource healthcare settings still rely on paper medical records. A host of barriers, from lack of required network infrastructure to concerns around staff adoption, ensure that the majority of health data remains trapped on paper. Without the ability to digitize this information, these providers are unable to perform disease surveillance, improve resource allocation, and gain insight from past outcomes to help communities that need it most.
For clinics and hospitals stuck in the gap between needing healthcare informatics and having the resources necessary to support a full EMR system, we developed a novel digitization tool capable of automatically extracting vital information from paper medical records. The software uses computer vision to digitize paper records without the need to re-design existing records. The tool is engineered to integrate into existing medical informatics workflows, and enable sustainable data acquisition where it was previously impossible.
Scan, upload, and annotate your own forms as they are. No need to re-design existing records.
AI software parses checkboxes, multiple-choice responses, handwritten digits and characters.
No reliance on internet or network infrastructure, just a laptop and scanner or high-quality camera
Allows transcription of free-form data, such as unstructured notes
At the Mae Tao Clinic this challenge of data acquisition is readily apparent. Operating for 30 years in the middle of the world's longest-running civil war, the Mae Tao Clinic offers essential health care services to migrants and refugees along the Thai-Burma border. In their setting, existing electronic medical record systems prove infeasible. And yet the clinic has a clear need for medical informatics - from cross-border disease surveillance to population health monitoring, access to health data is critical to the clinic's goal of supporting the well-being of 100,000+ migrants living outside the official healthcare system. Although it now offers a breadth of health-related services, its specialty has always been reproductive health. In the last 10 years alone the clinic has supported over 33,000 successful deliveries and provided over 136,000 antenatal care consultations. This record is all the more impressive given that the nearby region of eastern Burma displays some of the worst maternal health statistics in the world.
The story of Mae Tao Clinic's impressive record of care is perhaps best told by the data in their Health Information System (HIS). This homemade system contains an account of every patient visit at the clinic in the last decade, including over 1.1 million unique registrations. However, the need for manual data entry means that only a small subset of medical information contained on paper records can be tracked in the system.
For example, preliminary analysis of mothers who delivered at MTC revealed that while the average number of total antenatal care visits increased over the last decade, there was no clear improvement in access to first trimester antenatal care. At the same time, mothers who did not come to the clinic for first trimester antenatal care were 2x as likely to deliver a baby with low birth weight, and 1.5x more likely to have a premature delivery. Automated digitization would allow Mae Tao Clinic to track the full set of obstetric risks noted on the clinic's antenatal record, allowing us to uncover the more specific drivers of such outcomes, and suggest clinical changes to improve quality of care.
Since June 2019 we've worked with Mae Tao Clinic's Reproductive Health Department to integrate automated digitization into their existing medical record workflow. Above all else we ensure that digitization does not interfere with care or become burdensome to clinical staff. On-site testing begain in December 2019 and the intial results are highly encouraging! Our recently published paper used digitized data to demonstrate the relationship between antenatal care access and maternal health outcomes among mothers who deliver at the clinic, and has been used to advocate for additional funding amidst the ongoing migrant and refugee crisis in the region.
Starting in 2022 we began collaboration with Mbuya Nehanda Maternity Hospital within the Parirenyatwa Group of Hospitals in Zimbabwe to integrate autmated digitization into the data collection efforts in their neonatal ward. The project is part of a broader effort from the African Neonatal Association to assess neonatal outcomes across the sub-Saharan region, where quality improvement and health surveillance efforts have been historically limited by lack of population-level data. This work is supported by the Brown Global Health Initiative and Harvard Innovation Labs.
Step 1
Clinicians perform an antenatal care consultation or assist on a delivery. All information, including relevant history, noted obstetric risks, and outcomes, are recorded on the existing paper forms as usual.
Step 2
The clinician captures an image of the record once it has been completely filled out. Options for image capture include a smartphone, flatbed scanner, mounted camera, or even a high-quality webcam.
Step 3
Once the image is captured, the automated digitization process is performed in a matter of seconds. At this point the clinician has the opportunity to review the extracted data and alter or correct any information.
Step 4
When the clinician is happy with the results, they simply hit Save to add the new record to the system. The data is stored in a structured format that is compatible with the clinic's existing Health Information System, allowing community health workers, clinic managers, and partner organizations to analyze trends in real time or retrospectively. After saving, the paper record can be stored and accessed as usual.
We see automated digitization as part of a solution for providers in low-resource healthcare settings to collect and leverage data when EMR is not feasible. By getting data off the page, digitization enables everything from predictive healthcare analytics and resource optimization to data-driven research into the drivers of healthcare outcomes in remote populations. Possible extension of our tool include:
Digitization directly following point-of-care would allow the system to provide immediate feedback to front line medics. For example, the system might alert a medic that a mother's blood pressure and weight should be considered obstetric risks worth addressing.
Initial analysis of MTC's HIS validated some previous held assumptions about ANC care and delivery outcomes. With more robust patient-level information, we hope to better understand the specific obstetric risks driving outcomes in MTC's patient population and use this information to improve quality of care.
Ultimately we see our tool as part of a broader solution for medical informatics in developing healthcare settings. Integration with the right data storage and analysis tools would provide an all-in-one platform for not only data acquisition but also maintenance and analysis.
MPH, Harvard School of Public Health
PhD Candidate, Stanford University
Product Designer, Rippling
Associate Professor, Brown University
Founder of OpenMRS
Founder and Director
Mae Tao Clinic
Deputy Director of Medical Services
Mae Tao Clinic
Health Information System Manager
Mae Tao Clinic
Director
Brown Center for Biomedical Informatics
The project was made possible by funding from the Brown Data Science Initiative, Brown Global Health Initiative, and private donors. We are thankful for mentorship and support from Harvard Innovation Labs, the team at OpenMRS, faculty at the Brown Center for Biomedical Informatics and, most of all, our partners in Thailand and Zimbabwe who motivate us to continue this work.
