Reviewing Studies and Comparisons
There continues to be ongoing studies comparing the different biometric characteristics and modalities and their use in various industries including in healthcare and public health programs. Reviewing these resources can help when deciding which biometric characteristic(s) and modality(ies) to implement.
For example, Simprints published and shared their A Responsible Biometric Deployment Handbook, which provides a comparison and factors to consider for evaluation for select biometric characteristics and modalities.18 The table below builds on the work Simprints presents with the addition of links to examples of in healthcare and public health programs. These links include a selection of research and new articles, blogs, and/or manufacturers information.
Unimodal
Face
âTypically measures geometry and relationship of facial features.â
Ubiquity/Maturity
âWidely implemented and understood in commercial applications, with a variety of consumer and other applications.
Relatively well understood, and widely deployed for law enforcement and other applications.â
Drawbacks
âOverlaps with Law Enforcement and may make Facial Recognition challenging to obtain acceptance for in some settings.â
âIn cultures in which facial covering has a cultural or religious aspect, facial recognition may not be appropriate, and introduce gendered challenges both in terms of user and subject.â
âMost conducive to âtrivialâ reuse â e.g. users may immediately be able to reuse pictures of users to recognise or identify them.â
âWhile a robust understanding of accuracy exists in the academic community, and issues of bias are relatively well understood, they are not always solved or resolved.â
Benefits
âContactless â posing fewest public health challenges (e.g. transmission of surface or airborne disease).â
âNo specialist hardware required, making facial recognition relatively cost-effective.â
âWhile Liveness Detection is not embedded into all solutions it is widely available.â
âSome protective techniques such as tokenisation or encryption are available to protect facial templates.â
Examples and/or Studies in Healthcare and Public Health Programs
Ghana
Simprints, Gavi, Arm and Ghana Health Services
Thailand
NEC Thailand and Siriraj Piyamaharajkarun Hospital
United States
Geisinger working with CERTIFY Health
Fingerprint
âEssentially a digital analogue of the analogue process undertaken in law enforcement, using the scientifically established near-uniqueness of fingerprint features to identify humans.â
Ubiquity/Maturity
âWidely implemented and understood in commercial applications, with historic usage across law enforcement and security applications.
ISO standardised, making some Fingerprint templates interoperable between solutions.â
Drawbacks
âThe overlap with law enforcement may make fingerprint systems less acceptable in some contexts.â
âMost fingerprint systems require specialist hardware, increasing cost.â
âGenerally contact-based, increasing the risk of surface-borne disease.â
âSome user training is required to use hardware effectively.â
âMay not work with seriously physically impaired subjects.â
Benefits
âSignificant understanding exists regarding bias and accuracy, and whilst skin colour can be a confounding factor, the use of hardware which controls the capture environment offers some ability to mitigate this.â
âOf available solutions, fingerprint may be the most interoperable.â
âFingerprint recognition also benefits from some of the most sophisticated techniques for template protection and encryption.â
âFingerprint recognition captures relatively little âextraneousâ information (e.g. medical conditions etc) once the template has been captured.â
âWhilst not unspoofable, the need for specialist hardware makes replay or spoofing attacks harder to carry out, increasing resistance to fraud.â
Examples and/or Studies in Healthcare And Public Health Programs
Nigeria
PHIS3
Haiti
I-TECH (DIGI) - University of Washington (UW)
CĂ´te dâIvoire
CDC Contractor and SEJEN
Ethiopia
Ministry of Health - Ethiopia
Nepal
Simprints
Ghana
Kintampo Health Research Centre
Brazil
NatoSafe and Center for Womenâs Health (CAISM) at the University of Campinas (UNICAMP)
Mexico
Palm Geometry
âMeasures the shape and size of fingers using camera or similar technology.â
Ubiquity/Maturity
âLess common, but a number of solutions are available. Palm geometry has been in use in the private sector (e.g. banking) for some time.â
Drawbacks
âLess interoperable and understood.â
âHistorically, this has required specialist hardware, although can now be undertaken via camera / with commodity hardware.â
âFewer techniques available for sophisticated protection e.g. using template protection or tokenisation schemesâ
Benefits
âRelatively high social acceptance - captures minimal âextraneousâ information such as medical conditions.â
âContactless - posing fewest public health challenges (e.g. transmission of surface or airborne disease).â
âNo specialist hardware required, making facial recognition relatively cost-effective.â
âPalm geometry captures relatively little âextraneousâ information (e.g. medical conditions etc) once the template has been captured.â
Examples and/or Studies in Healthcare And Public Health Programs
Company Tools/Products/Reviews
Iris
âMeasures the eye itself, recording and comparing unique features in the iris.â
Ubiquity/Maturity
âLess common, but subject to some large-scale deployment in humanitarian settings.
Iris recognition has been deployed in the private sector (e.g. banking) for some time.â
Drawbacks
âLess interoperable and understood.â
âHistorically, this has required specialist hardware, and specific capture conditions - and requires proximity to capture, increasing cost and complexity at point of use.â
âMay not work with seriously physically impaired subjects or individuals subject to eye surgery.â
âAffected by lighting change, and requires configuration and supervision.â
âLess inherently resistant to spoofing / impersonation attacks.â
âSome protective techniques such as tokenisation or encryption are available to protect facial templates - but fewer than fingerprint or face.â
Benefits
âRelatively high social acceptance - captures some âextraneousâ information such as medical conditions, but less than other modalities.â
âLike fingerprint, while not 100% unique, Iris patterning is extremely random and determined prior to birth, potentially giving rise to a very low false match rate.â
âWhile proximity is required, Iris is less âhigh contactâ than modalities such as fingerprint, posing reduced disease transmission risk.â
âWorks with cohorts of users who use facial coverings.â
âThe Iris is protected and less susceptible to damage than finger or hand.â
Examples and/or Studies in Healthcare And Public Health Programs
Kenya
United States
Norvant Health
Company Tools/Products/Reviews
Periocular
Captures âthe area around the eye, including eyebrow and other facial features above the mouth.â
Ubiquity/Maturity
âLess common; some recent research, but relatively few commercial products.â
Drawbacks
âLess interoperable and understood.â
âMay be confused by users with facial recognition.â
âFewer techniques available for sophisticated protection e.g. using template protection or tokenisation schemes.â
Benefits
âMay leverage consumer hardware and therefore reduce cost.â
âHigher social acceptance than full-face.â
âContactless â reducing surface-based transmission risk.â
âMay combine with Iris recognition to produce some of the benefits of both systems.â
Examples and/or Studies in Healthcare And Public Health Programs
United States
Palm Vein
Captures âthe sub-surface veins in hands, typically using infra-red light and a specialist sensor.â
Ubiquity/Maturity
âSome recent research but less common. Fewer commercial products but a number of actively innovating projects and vendors.â
Drawbacks
âLess interoperable and understood.â
âRequires specialist hardware - increasing cost and reducing interoperability.â
âStill in active innovation - subject to change over time.â
âFewer techniques available for sophisticated protection e.g. using template protection or tokenisation schemes.â
Benefits
âHigh social acceptance.
âSome medical data is captured, but relatively little extraneous data (e.g. facial image) which can trivially be reused.â
âContactless - reducing surface-based transmission risk.â
âVein patterns are relatively unaffected by age, disease, or physical damage, increasing accuracy over time.â
Examples and/or Studies in Healthcare And Public Health Programs
United States
University of Utah Health Care
BayCare Health
Voice Recognition
Captures âthe voice using an audio sensor.â
Ubiquity/Maturity
âRelatively wide use, but fewer products targeting users in the Global South.â
Drawbacks
âLess interoperable.â
âMay be more susceptible to confounding accuracy factors with populations with languages and dialects who solutions have not been designed for.â
âFewer techniques available for sophisticated protection e.g. using template protection or tokenisation schemes.â
Benefits
âHigher social acceptance than other solutions.â
âRelatively no extraneous data is captured.â
âContactless - eliminating surface or airborne transmission risk.â
âLeverages widely available commercial hardware.â
Multimodal
âCombination of multiple techniques or modalities.â
Ubiquity/Maturity
âLess widely deployed, but a number of products are integrating multimodal support. There are some systems at large scale using multi-modal biometrics.â
Drawbacks
âDepends on the schemes used.â
Benefits
âDepends on the schemes used - but potentially multimodal biometrics presents the opportunity to combine and trade benefits and disadvantages of multiple schemes.â
Ethiopia â Iris and Fingerprints
Senegal â Face and Fingerprints
Face and Gestures
Members from the Data Use Community shared chosen biometric systems, and insights learned when researching biometric systems to implement.
For example, as mentioned previously, Nigeria, CĂ´te dâIvoire, Haiti, and Ethiopiaâs eCHIS implement unimodal biometric systems that utilize fingerprint biometrics.13,14,15,16 Ethiopia is looking to implement a multimodal biometric with fingerprint biometrics and iris biometrics for the national ID16, and Simprints worked with countries utilizing unimodal and multimodal biometrics systems.14
In addition, after setting up a patient identification management system, Zimbabwe identified implementation of a biometric system as a next step. Zimbabwe shared considerations and comparisons from their research of the use of fingerprint biometrics and/or iris biometrics. Their findings yielded the following advantages and disadvantages of biometric system implementation in Zimbabwe: advantages - improving patient identification, time savings, accessibility, scalability, and a prototype already being developed; and disadvantages - implementation and maintenance costs, and infrastructure limitations.19
Last updated