ScanForm

Timely and accurate data is a critical need around the world, but is hindered by inadequate infrastructure and human resources. Only 28% of health facilities in Sub-Saharan Africa have access to reliable electricity, making all-electronic data collection infeasible in many places [1]. 

Facilities that are not electrified depend on paper-based data collection. Challenges with traditional paper include manual ferrying, manual transcription, and manual summarization, creating a cascade of inaccuracies from human errors. This process is also time consuming, causing delays for database reporting and analysis. 

In facilities with Electronic Medical Records (EMRs), power and network outages force a return to paper and create a backlog. Back-entry from paper is slow and very incomplete, and further burdens healthcare workers because when power/network returns, retrospective data collected on paper must be entered while new clients await services.  

The development of PHCPI’s 38 Core Indicators provides critical guidance for countries to standardize the evaluation of primary healthcare performance, but this data is obtained from third party sources with questionable quality [2]. Whether paper-based or electronic, these sources are negatively impacted by the aforementioned challenges.  

A simple, interoperable solution is needed that allows facilities to quickly and accurately collect primary healthcare data, without reliable electricity or network. By strengthening the paper sources that generate the PHCPI Core Indicators, we address the root of the challenge with measuring primary healthcare improvement.

ScanForm is a non-disruptive solution that transcribes handwriting on paper forms into digital data with a single smartphone photo. Once the photo is uploaded, artificial intelligence (AI) extracts all information to auto-calculate custom summary statistics and visual analytics in under 60 seconds, saving time and eliminating human errors. If the network is unavailable, data collection can still occur offline by taking photos, which are queued for automatic upload once the internet is restored.  

ScanForm is complementary to existing methods of data collection in low- and middle-income countries (LMIC) because pen/paper is familiar and the level of effort to record information stays the same. In collaboration with local governments and organizations, paper tools and source registers are redesigned and printed to be compatible with ScanForm, ensuring all original data elements are captured. In addition to data collection, ScanForm also handles the end-to-end processing of data, including: cleaning, summary statistics, data quality assessments, imputation, and phone management, all custom built to your specifications. As a result, ScanForm delivers comprehensive confidence in data quality and reduces multi-month data processing cycles to near real-time. 

Simple and scalable, ScanForm is a better way to collect paper-based source data because of the improvement in data quality from automatic transcription into digital data and the elimination of human errors. Data collection with ScanForm is at least 3 times faster compared to tablets, saving valuable time and effort for healthcare workers. Summaries are delivered to smartphones via the ScanForm app, empowering healthcare workers to review the results they collected from their facilities, and eliminating millions of hours of manual summary calculations at scale.

Additionally, ScanForm is interoperable with other systems like EMRs and DHIS2. By merging with electronic sources, ScanForm can create a complete picture of primary healthcare performance from paper and electronic-based methods of data collection. 

In health applications, ScanForm has been successfully used by clinicians, researchers, healthcare workers and community volunteers for: epidemiological surveillance (HIV, malaria, COVID-19), routine health reporting, clinical trials, and influenza vaccine delivery. With extensive experience in facilities and communities, ScanForm is suitable to collect paper-based primary health care data by anyone already doing so.

ScanForm serves anyone collecting data on paper, particularly healthcare providers and community volunteers. These dedicated frontline workers are committed to improving the health of their communities, despite facing incredible challenges with paper-based reporting, coupled with untenably low wages in developing countries. They are also the first to be burdened when external donor reporting requirements add new indicators and disaggregates. This data is the necessary foundation for new insights and understanding, but is often collected at the expense of those who are least equipped to handle increasing reporting complexities from areas with unreliable infrastructure.    

The purpose of ScanForm is not just to alleviate the workloads of overburdened healthcare workers, but also strengthen the evidence of their collective impact by ensuring their carefully recorded data are received quickly and accurately. Oftentimes, data flows in one direction: from the primary collectors up to higher levels. In this way, healthcare workers and facilities are underserved because summaries or conclusions from the data they report are usually never sent back.  

Data is information and information is power. In addition to the digitalization of handwritten data, ScanForm produces comprehensive confidence in data quality by automatically measuring accuracy, completeness and timeliness with each image upload. Auto-calculated summaries of the collected data, along with automated data quality assessments, are delivered via the ScanForm app, ensuring that critical and reliable information about primary healthcare performance is accessible. By returning detailed information back to where it was collected, ScanForm empowers healthcare workers with accessible insights to inform performance improvement efforts at their facilities and communities. 

Quantitative Engineering Design (QED.ai) is a technology company that builds data systems and artificial intelligence to empower organizations pursuing the Sustainable Development Goals in health, agriculture, education, and justice. We are fully mission-driven with expertise in development and humanitarian projects in Sub-Saharan Africa and South Asia. By leveraging our technologies and knowledge, we shape and adapt software and hardware to address needs on the ground, to better collect and analyze data to inform policy decisions. 

With offices and local staff in Malawi, Kenya, Taiwan, Poland, and the United States, having a strong local presence is a priority at QED. Our technologists collaborate with organizations by working with local experts and embedding with them for extended periods of time. Together, we develop practical solutions informed by realistic requirements and infrastructural limitations. Training is a crucial part of our work, to develop local capacity and ensure sustainability. 

ScanForm is suitable for any paper tool and designed for challenging environments, based on realistic experiences in rural communities. When new scannable forms are made, a prototype is first developed based on paper tools currently in use, preserving all data elements in their original order often as dictated by local Ministries of Health. Multiple iterations are then made based on feedback from focus groups, until all expectations are satisfied. Our forms and training materials are also adapted to local languages, such as Swahili in Tanzania, and French in Benin. By constantly incorporating feedback from users, and working in collaboration with local governments, ScanForm’s improvement is guided by the inputs and ideas of the community.

We are applying to this Challenge because we have a cost-effective, timely and accurate solution to measure the improvement of primary health care performance in LMIC. A paper questionnaire that is scannable with ScanForm can be disseminated to any facility to measure the PHCPI Core Indicators. After completing the form and taking a picture, the results are digitized and can be merged with any data from electronic sources for a complete evaluation of primary healthcare performance. The results can be sent back to the facility and community health assistants via the ScanForm app, empowering healthcare workers with accessible insights. When implemented routinely, primary healthcare performance can be longitudinally tracked to monitor improvement over time. 

As ScanForm continues to expand to more countries, we are applying to this Challenge to help us overcome financial and market barriers. Increased funds would facilitate faster research and development for better human-centric design and supporting additional languages, plus infrastructure investments for more in-country deployments and unlocking stronger economies of scale. We also hope the Challenge can increase awareness of ScanForm through MIT’s and BMGF’s networks and its vast potential of applications, since drawbacks with paper-based reporting are common. Many competitors in the digital health market believe EMRs can work everywhere, which is not the reality yet, particularly in the Global South. Until then, digital data can be captured from paper-compatible ScanForms and merged with electronic systems. 

ScanForm is a game changer for data collection in global health and beyond. By leveraging the state-of-the-art in artificial intelligence (AI) and software engineering, challenges with paper-based methods of data collection are solved. ScanForm is innovative because its AI is custom-calibrated to local handwriting from the countries where it is deployed. The optical character recognition (OCR) technology is very accurate (>99%) and self-learning, such that ScanForm continuously improves to best fit its environment. This contrasts with many alternatives in the global health space, which require scanners and the physical removal of forms from their original locations, and OCR tech that is not self-learning and has not improved since the 1990s.

As a data system, ScanForm can function independently in areas which lack the IT infrastructure to support fully electronic data capture. Or, ScanForm can work as a transition platform for countries scaling up their IT infrastructure. Once that infrastructure is ready, ScanForm data can integrate harmoniously with electronic systems to achieve a holistic national digital health data system. This interoperability makes ScanForm catalytic, because it can succeed where electronic solutions cannot.   

ScanForm is also innovative in the speed with which it closes the feedback loop between data entry and data analysis. Our value propositions here include:

• Summary statistics and data quality assessments (DQAs) are generated within seconds, even though the data is recorded on paper. 

• Limitless disaggregating of paper-based data, without extra human labor or human errors.

• Fully remote DQAs dramatically reduce the costs of travel and per diems.

• Smart imputation enables automatic improvement in data completeness without additional time or effort from healthcare workers.

• ScanForm is also interoperable with DHIS2 and EMRs to ensure that data and analytics are sent to approved national repositories.

In the next year, we believe the existing evidence of ScanForm successes will lead to future opportunities to assist new countries with strengthening their data collection processes, particularly for LMICs in Sub-Saharan Africa. Through collaborations with new ministries of health, we will redevelop scannable routine health registers to ensure that data on paper is captured quickly and accurately. Within the coming year, the full national scale-up of ScanForm will be completed in Malawi. We anticipate a continuation of our impacts, saving money by eliminating costs, reducing reporting burdens, and improving data quality. These benefits can apply in any situation where paper is depended on because the infrastructure for electronic solutions is unreliable or nonexistent.  

Over the next five years, ScanForm may become a core technology used in multiple national health systems. This will be achieved by improving the completeness and quality of data collected from paper, which will be analyzed faster to critically inform public policy and safety. Over this time, we anticipate the development of more significant features for ScanForm. Taken altogether, ScanForm will be the backbone of paper-based data collection and continue to improve based on its customization to where it is used. In the next five years, we expect more than 50 million patient data records to be processed with ScanForm.    

• Number of active collaborations with ministries of health and international organizations

• Number of government registers and data collection tools compatible with ScanForm 

• Number of working days in the month saved for frontline healthcare workers, by eliminating duplicative reporting and transcription

• Number of key performance indicators required by health departments, donors, and external stakeholders that ScanForm can auto-calculate 

We expect our solution to have an impact because it addresses multiple challenges with paper-based data collection, which is the norm throughout LMIC. By implementing the logic model below, we can ensure that primary health care data is captured well — timely, accurately, and completely. This higher quality of primary data is necessary to understand what gaps exist to inform performance improvement efforts. 

Activities:

• Redesigning an existing paper tool (or creating a new tool) for compatibility with ScanForm to measure PHCPI Core Indicators 

• User acceptance testing of prototype, reiterations based on feedback

• Interoperability with existing electronic systems, if applicable 

• Training(s) with facility staff on how to collect data with ScanForm tool and how to use ScanForm app  

Output:

• Increased human capacity to use ScanForm 

• Distribution of a scannable paper tool for digital data collection at facilities in LMIC to measure PHCPI Core Indicators 

• Auto-calculated indicators and summaries, without human errors 

• Results from data quality assessments, for confidence in data quality 

Outcomes:

• Transparent and reliable results without manual data entry

• Improved data quality via increased accuracy, completeness, timeliness

• Summaries and data quality reports delivered back to the facility

• Increased efforts by healthcare workers to address gaps in primary health care performance at the facility  

• Quantifiable improvement in primary health care performance at the facility that can be tracked longitudinally

ScanForm’s core technology is an end-to-end pipeline for the accurate capture and downstream processing of handwritten data. This pipeline begins with paper forms, intelligently designed to present questions crisply and compactly. These forms are enhanced with metadata about data types, validation checks, and security constraints. Photographs of handwritten forms are then captured offline by an Android mobile application, which automatically uploads to back-end servers once internet is available. (This infrastructure can be either cloud-based or deployed on-site, depending on needs, regulations, and environmental constraints.) The photographs are processed using custom-built deep learning networks calibrated to local handwriting and with self-learning capabilities. The data is then automatically cleaned, reshaped, imputed, and transformed into near real-time analytical reports to meet the needs of ministries of health and other stakeholders. The entire pipeline runs within seconds.

Healthcare workers and community health volunteers collect primary health care data with ScanForm. The vast majority are employed by the government, so collecting this data is a routine part of their duties. Upgrading routine registers and tools to ScanForm is non-disruptive because it leverages universal familiarity with paper-based data collection. These processes are now enhanced by automatic transcription and summarization, without manual effort or a cascade of human errors. Frontline healthcare workers are thus incentivized by very significant time savings, regaining anywhere from 2 to 10 working days per month that would have otherwise been spent on rote transcription and hand calculations.

Evidence from comprehensive focus group discussions in Malawi show very high user acceptance of ScanForm. Healthcare workers overwhelmingly reported that ScanForm saves time (95%), and 100% responded that data collection is faster with ScanForm. Satisfaction and comfortability were also reviewed highly, with 98% of healthcare workers stating they would be somewhat or very likely to recommend ScanForm for their facilities. Overall, 95% of participants rated the technology as good or very good, further indicating ScanForm is a suitable solution that users want to use.  

QED is committed to endorsing diversity, equity and inclusion. Over 50% of QED’s staff are female. It is a minority and women-owned business, with two Asian American directors. QED also establishes local teams and offices in many countries where ScanForm is used, hiring local personnel to lead deployment and capacity building efforts, with directors also living in East African countries for years (e.g. Malawi, Kenya, and Tanzania). In sourcing talent, QED is often interviewing and empowering community health volunteers, who are often the ones responsible for running both health facility services and community health programs, as previously unpaid laborers. Our European branch employs Ukrainian staff and is actively assisting the Ukrainian refugee diaspora.

QED also works in direct collaboration with governments to ensure that their points of view are incorporated into the data tools that are built together with them. Lastly, QED has a zero tolerance policy for racial discrimination, age discrimination, or sexual harassment, and has exercised same-day terminations when these rare events have occurred.

Our primary business model is software as a service (SaaS) for our customers, which are ministries of health (MOH) and affiliated organizations. In addition to data capture and processing, our services often include all printing, training, hosting, phone procurement and management, and technical support. We provide these services virtually and with our on-the-ground teams in countries where ScanForm is deployed. In the first year, establishing data pipelines and standards, making procurements, and running trainings and sensitizations constitute the bulk of one-time initial costs. Implementation costs decrease substantially afterwards, as maintenance is inexpensive once these pipelines and human capacities are established. From our investigations, the ScanForm approach is at least two orders of magnitude cheaper than traditional approaches which attempt to deploy EMRs to all points of care. 

ScanForm addresses pain points for our clients in public health because it eliminates many costs currently associated with manual calculations, data quality assessments, and IT maintenance. ScanForm is much more sustainable because we train MOH staff to take full control of the data entry process, and no auxiliary staff or special equipment is required. By adopting this business model, countries stand to not only improve their data quality but also financially benefit from using ScanForm at scale.

To date, ScanForm has been financed by grants, proposals, and personal resources, and has never received investment funding. In the long term, we expect more governments throughout the Global South to establish multi-year service contracts with QED to use ScanForm at national scale. 

The value proposition for these Ministries of Health is that ScanForm is a sustainable solution, easy-to-use, which can streamline and dramatically reduce the long-term costs of data processing for many epidemics, including HIV, malaria, COVID-19, and integrated disease surveillance. 

QED has received support from organizations such as CDC, PMI, WHO, BMGF, and the Global Fund, to deploy ScanForm in various African regions, including Kenya, Malawi, Burkina Faso, Benin, Namibia, and Tanzania. The largest deployments to date have occurred in Kenya and Malawi.