Kenya varsity successfully tests facial recognition system for HIV patients

By Gatonye Gathura

Moi University, Kenya, has tested a face recognition system they report to have a 99 per cent success in patient identification.

A test run carried out within the US-funded HIV Ampath project in western Kenya reports a 99.3 per cent success in patient recognition.

The success is reported at a time Kenya government is under pressure from the US to speed up the operationalization of Huduma Numba.

Last year the government initiated the National Integrated Identity Management System (NIIMS) otherwise called Huduma Namba as the ‘single source of truth’ on personal identity.

Impersonation

The number which is yet to be operationalized was also meant to reduce cases of fraud and impersonation in public service access and delivery.

Since 2016, the US President’s Emergency Plan for AIDS Relief (PEPFAR) has been pushing the Ministry of Health to introduce a unique patient identifier system in its HIV projects to reduce cases of theft, commodity loses, and data manipulation.

The introduction of such a system is one of 15 requirements Kenya must meet by October or risk losing HIV funding from the US.

“For COP 2020, the failure to meet any of these requirements will result in reductions to the Kenya budget,” says Dr Deborah L. Birx, the US HIV coordinator in a letter to the American ambassador to Kenya.

Huduma Namba

In the letter, written earlier this year Dr Brix asked Ambassador Kyle Mccarter to liaise with Kenya on the operationalization of Huduma Namba.

“There is need to have the long timeline on the adoption of huduma number addressed,” said Dr Birx.

The COP 2020, for which the US has budgeted about Sh 40 billion ($375,000,000) commences in October to the same time in 2021.

However with current disruptions from Covid 19, Kenya is unlikely to meet any of the PEPFAR minimums including the operationalization of Huduma Namba.

Ethical

But while the Maseno University invention is a success, the authors say a lot of work still remains to be done because so far it has only been tested on a small population.

The technology, they say will also require rigorous evaluation of the ethical, cultural and regal requirements and acceptability before it can be applied.

The study also involving Vanderbilt University, US and Uganda Christian University,   had recruited 103 patients from the Academic Model Providing Access to Healthcare (AMPATH) program which provides care to 95,000 HIV patients in western Kenya.

AMPATH is a partnership between Moi University, Moi Teaching and Referral Hospital, the Ministry of Health and a consortium of about 10 North American universities.

The team led by Sight Ampamya of the Institute of Biomedical Informatics, Moi University, says the use of unique identifiers can also reduce patient identification errors in hospitals.

“Identification errors are not uncommon as was evident in March 2018 at Kenyatta hospital when a brain surgery was done on the wrong patient.”

Poor Countries

The aim of the study published last month in the International Journal of Medical Informatics, was to investigate whether a facial recognition technology can work in poor countries such as Kenya.

Participants, all adults had their facial images captured, processed and used to train the computer intelligence system.

Then at another station within the hospital the patients pretended to be revisiting the clinic where their facial images were re-captured.

“Of the 103 patients, 102 were correctly identified on the first try, for a sensitivity of 99.03 per cent,” says the study.

The one participant who was not identified on the first try, the authors say was identified correctly on the second attempt.

Spectacles

Wearing of spectacles did not affect performance, with all the seven patients wearing spectacles correctly identified

“Our study offers a potential solution to the problem of unique patient identification in poor countries,” concludes the study.

However, the innovators say the system will need to be tested further with larger patient populations to assess the processing speeds as the number of images increases.

But also important, they say is to make sure the facial recognition is ethically and legally acceptable before it can be put to public use.

In November President Uhuru Kenyatta signed the Data Protection Act 2019 which regulates the collection, processing, and use of personal data in Kenya.

Biometrics

While the law on data regulation is now in place it still might take time to establish the office of the Data Protection Commissioner who will oversee the operationalization of the Act including the formulation of relevant policies and regulations.

The use of biometrics for patient identification in HIV programmes in Kenya has in the past met strong opposition from the civil society

In 2016 the ministry had to abandon a planned national registration of gays, sex workers, and drug injectors despite having acquired expensive biometrics data capture equipment.

The ministry wanted to develop a foolproof method for identifying individuals collecting free HIV products, services, and payments.

But civil groups had argued the plan was illegal and that such data could be misused to identify and expose participating individuals.

About Gatonye Gathura 40 Articles
Science Journalist

Be the first to comment

Leave a Reply

Your email address will not be published.