Updated Oct 03, 2019

Simprints Biometric Identification System


Christine Kim

Send Message

Stage 5: Scaling

We currently have >90 client projects lined up in our pipeline and 3 currently deployed in the field. Last August, we successfully equipped Dimagi’s CommCare software technology with biometric capability so that the health workers of Possible in remote Western Nepal will be better able to identify over 5,000 patients during maternal health care visits and ensure they receive vital follow-up care. As of March 2018, we will be expanding with Possible to the Dolakha region, totaling up to 50,000 beneficiaries reached. In October 2017, we deployed in Dhaka, Bangladesh with BRAC’s Maternal Health Program (Manoshi). Working with 45 of their community health workers, we are reaching 22,000 mothers and children who will receive maternal and child health services through BRAC's ehealth platform, newly-equipped with Simprints' biometric identification system. In February 2018, we partnered with Watsi in Rwibaale, Uganda, enrolling 6,000 villagers into their low-cost healthcare programme. By using biometrics to track primary care expenditure, we aim to validate an innovative and sustainable healthcare model in one of the world’s toughest environments.
We implemented the Simprints biometric identification system with Possible Health, our nonprofit healthcare partner in Achham, Nepal. 19 community health workers are currently using our system to deliver more accurate and inclusive healthcare to over 30,000 patients in surrounding villages. We deployed in Dhaka, Bangladesh withBRAC’s Maternal Health Program (Manoshi). Working with 45 of their community health workers, we are reaching 22,000 mothers and children who will receive maternal and child health services through BRAC's ehealth platform, newly-equipped with Simprints' biometric identification system. Watsi in Uganda is our latest partner to use Simprints. 6,000 villagers in rural Rwibaale will be enrolled into their low-cost healthcare programme and identified at the local clinic using our biometric solution.
Our goal is to radically disrupt the inaccurate way we currently track and deliver progress towards the SDGs, and instead build a world where every person—not estimations and guesswork—actually counts in the fight against poverty. The beauty of biometrics is its inherent advantage of being able to measure the exact number of every beneficiary touched by our projects, and Simprints is committed to making robust impact measurement a core part of our strategy. To that effect, we have completed two rounds of field tests: the first comparing the matching accuracy of the Simprints scanner to five existing commercial biometric scanners in low-resource contexts (e.g., marked by a higher proportion of worn, burned, and scarred fingerprints, and extremes in humidity, heat, and dust) across Bangladesh, Zambia, and Benin; and the second testing our prototype biometric system in Achham, Nepal to assess whether its matching accuracy is sufficient for reliable beneficiary identification in these low-resource contexts. For the first round, conducted in 2015, the results were as follows: the Bangladesh cohort consisted of 33 participants in a maternal/neonatal community health clinic who collectively gave 3,917 fingerprint images. Most were slum dwellers, many of whom were garment workers. Using a second fingerprint brought the Simprints Equal Error Rate (EER) down to 0.00%. The Zambia cohort consisted of 428 participants, mainly lifelong farmers whose fingerprints were especially worn or calloused and therefore difficult to image. Altogether 68,162 fingerprint images were collected, and all sensors reflected high error rates. However, using a second fingerprint brought the Simprints EER down to 2.01%, which approached 0.00% when secondary tags were used. Finally, the Benin cohort was made up of 289 participants, 48 men and 241 women of mainly 20- 40 years of age employed in a broad range of suburban and rural jobs. Altogether, 50,119 fingerprint images were collected from the population that could be most broadly generalized to developing country contexts. Using a second fingerprint brought the Simprints EER down to 1.57%. In conclusion, we found that of all the sensors in all settings in which they were tested, the Simprints scanner consistently outperformed the five commercial products. See supporting documents for details. In 2016, the second round of field tests was carried out in three villages in Achham, Nepal. The Simprints scanner was used to collect right and left index and thumb fingerprint images from 123 individuals: four impressions of each finger were collected in an enrollment and a verification session, separated by 2-3 days. The enrolled cohort was largely composed of female farmers, with almost 1 in 10 fingers showing significant signs of damage. Depending on how many fingers (1-4) were used to identify a beneficiary, the mean matching scores ranged from 86.99% (right-hand thumb only) to 95.12% (right and left index and thumb). While highest true positive identification rates occurred when including all fingers fingerprinted, high accuracy identification was still achieved with most combinations of two fingers. In conclusion, the results suggest that using the Simprints biometric identification system to collect fingerprint images for enrollment and identification will achieve high accuracy to usefully identify beneficiaries in the field, reduce record duplication, and minimise the risk of exclusion. See supporting documents for details.

Registered in United Kingdom.

Focus Areas:

Agriculture, Democracy, Human Rights, and Governance, Education and 10 MoreSEE ALL

Agriculture, Democracy, Human Rights, and Governance, Education, Health, Humanitarian Assistance, Water Sanitation and Hygiene (WASH), Youth, Digital Finance, Digital Development, Digital Inclusion & Connectivity, Human Centered Design, Public-Private Partnerships and SecuritySEE LESS

Implemented In:

Bangladesh, Kenya, Nepal and 2 MoreSEE ALL

Bangladesh, Kenya, Nepal, Uganda and ZambiaSEE LESS

Countries Implemented In
Funds Raised to Date
Verified Funding

Innovation Description

We build open-source software and biometric hardware that empowers the mobile tools used by NGOs, businesses, and governments fighting poverty around the globe. Recognising a significant gap in current biometric technology, our team has developed an open-source fingerprint system for the world’s toughest settings that is 228% more accurate and 4x cheaper than existing solutions. Our goal is to build a world where every person—and not guesswork—actually counts in the fight against poverty.
How does your innovation work?
Simprints is a non-profit tech company that seeks to break the identification bottleneck. We have built mobile biometric hardware and open-source software to empower the mobile platforms used by NGOs, businesses, and governments around the world. Our identification system is affordable, secure, rugged, and designed to work in the world’s toughest settings. To reach beneficiaries we partner with organizations that are already using or planning to use mobile services at scale. We do not believe creating yet another mobile development ‘app’ is the most effective route to impact. Instead, we are uncompromisingly open-source: we integrate low-cost, open-source biometric functionality into digital platforms that are already making—or poised to make—positive impacts in health, education, finance, and aid at scale. Simprints’ solution is the first ever biometric hardware and software system designed for—and with—frontline workers in developing countries. Our solution is comprised of three key components: a unique fingerprint scanner, an Android matching application (open-source and free), and an optional online Cloud/SaaS portal that offers project management, device syncing, and online de-duplication. This technology combines several key innovations essential to success in low-resource contexts. On the hardware side, we have achieved significant cost savings and mobility improvements by moving the biometric extraction algorithms that typically run on operating systems to the embedded software of our microchips. We have done this through a unique partnership with ARM Holdings, the world’s largest designer of microchips, which power 95% of all smartphones globally. This means we can transfer fingerprint templates (500 bytes) instead of full images directly from the scanner, an essential capability for transferring data in low-bandwidth settings. On the software front, our algorithms—unlike most biometric software—have been developed to be highly accurate with developing country fingerprints, particularly those that are worn, scarred, or burned. We have collected over 135,000 fingerprint images from Zambia, Benin, Nepal, and Bangladesh, leading the largest academic study of developing world fingerprints globally. With our sensor, every individual in this study was successfully registered, and 92.4% of fingerprint images were obtained correctly on the first attempt (<0.5s acquisition time). Our tests in health clinics achieved a 1.68% equal error rate (false accepts equal false rejections), which decreases to functionally zero when using multiple fingers and/or secondary tags, such as location. This results in a technology that is 4x cheaper and 228% more accurate in low-income contexts than existing tools. To ensure beneficiary privacy and security, our actual software databases never store biometric data alongside patient information. Instead, we use anonymous IDs to link between program data and our matching algorithm database. The system is highly secure, using 128-bit encryption between the scanner and the phone, and SSL/TLS encryption between the phone and the server. The server uses the OAuth 2.0 Standard with tokens generated from keys that are never passed over the web to authenticate and authorize requests. Fingerprint images are stored as encrypted numeric byte-array templates. In the unlikely event of unauthorized access, the nature of the templates makes it incredibly difficult to reverse-engineer the fingerprint image from an extracted image. We voluntarily adhere our data processes to the European Privacy Regulation standards, the strictest in the world. Altogether, these technical advances have allowed us to deploy projects that truly succeed in reaching the world’s poorest beneficiaries. For example, our latest project with BRAC, the world’s largest NGO, helps their community health workers reach over 22,000 mothers and children in Bangladeshi slums, of which 5,400 (25%) live in extreme poverty. Slums suffer from the pressure of high urban migration and frequent slum evictions that make keeping track of health records almost impossible. The very poorest mothers and children are most vulnerable to health shocks and the least likely to receive the antenatal and postnatal healthcare visits that detect, prevent, and treat issues before they become preventable deaths. In our Dhaka slum catchments, infant mortality rates are double the national average, and only 7.3% have access to government health clinics. Improving health services in these regions represents a highly cost effective and concentrated way to significantly improve SDG indicators. The Simprints solution is simple, powerful, and limitless in its potential for global impact. By bringing biometric identification functionality to mobile platforms, we can catalyze on the thousands of organizations around the world that are already making positive impacts in health, education, financial inclusion, and aid at scale. Together, we can ensure a world in which every person counts.

Competitive Advantage

All biometrics - iris, face, finger, vein, and even ear - require two components: hardware for image acquisition and software for matching. Originally, Simprints never intended to produce hardware or software. Our plan was simply to buy both and integrate them with digital platforms. However, we found proprietary industry biometrics were deeply inadequate for this context. Hardware All low-cost biometric scanners are wired by USB and designed for desktop use. Without an on-board battery, Bluetooth, and rugged casing, they are useless for mobile field applications in high humidity or dusty environments. Devices that are mobile have so far been designed for army, police, and security forces such as the FBI. These devices are extremely expensive and typically incorporate multiple functionalities such as on-board keypads, ID card readers, cameras, etc. that are superfluous to our core, price-sensitive use case. Software Biometric software companies will typically charge the clients per number of scanners used, per person enrolled in the system, per scan, as well as an annual maintenance fee or license. Cumulative costs can amount to more than $200,000 altogether. More importantly, no algorithm that we have tested handles deeply worn or burned fingerprints particularly well—common among the biometric profiles of rural farmers, or poor mothers holding hot pots with bare hands. Thus, while biometrics is a powerful way to connect beneficiaries to critical services, few nonprofits have integrated them into their workflows. We believe the bottleneck is the lack of a low-cost biometric scanner that requires minimal training, and intuitive software that can easily plug into existing platforms. Obstacles include high costs, integration complexity, training, and low interoperability between vendors. The current proprietary systems mean organizations are locked into closed biometric standards, reducing their ability to control their data, shape the system as they want, and collaborate with partners. These are exactly the obstacles that Simprints tackles in order to deliver impact. Our commitment to low-cost open-source software and rugged biometric hardware allows any project to utilize our system to solve their own problems, create improvements, and find applications that we can’t even imagine.

Planned Goals and Milestones

Simprints is an open-source biometric platform that can seamlessly plug into existing mobile and electronic medical record systems. Similar to widely scaled platforms like CommCare/ODK and OpenMRS, this technology will be sustained both by open-source contributors and Simprints. With over 30 LOIs, 10 actively paying clients, and >70 clients in a nascent sales pipeline, Simprints has a viable path to financial sustainability in a fingerprint biometrics market that’s growing at 18.9% CAGR and bound to reach $8.85B in 2022.
We envision three stages of growth. The first stage is R&D, which we have just completed after developing and validating this technology during our studies at the University of Cambridge. The second stage will be the testing phase (current location), where we complete critical software platform integrations and pilot our first projects with risk- friendly partners and early clients. The key objective will be to refine both our technology and our product-market fit, and we will manually undertake a lot of activities (e.g. purchases, training, and troubleshooting) that we intend to automate down the line. As the product and business model stabilize, and we develop a robust track record, we will shift to the third stage focused on commercial scale. We do not know yet whether large, slow government contracts or a high-volume of smaller NGOs and business clients is the most effective route to scale at this stage. Either way, we imagine a strong evidence-of-impact base and political support from policy shapers like the World Bank will be essential generating the confidence required to land huge projects. We are already actively building relationships at these institutions, and have been advising the World Bank’s Identity for Development (ID4D) group on biometric technologies.
Funding Goal8,000,000


Mar 2018
Date Unknown
New Country Implemented In
Bangladesh, Nepal, Kenya, Uganda and Zambia
Date Unknown
In the News
TITLEThe Economist: Getting to Cambridge: The political philosophy of Britain’s most successful city
Date Unknown
In the News
TITLEBusiness Weekly: Simprints Awarded Startup Company of the Year 2015

Supporting Materials

Simprints Nepal Report 2016 (2).pdf
Simprints Field Test Study Analyses_ Benin, Zambia, Bangladesh (2).pdf