NIGERIA —Researchers from the Australian National University (ANU) have developed the Hep B Live test, a machine learning tool aimed at improving outcomes in the diagnosis and treatment of Hepatitis B Virus (HBV) in Nigeria.
Using patient data from Nigeria, the team created an algorithm that learns from the data, identifies patterns, and makes intelligent decisions to provide alerts and detect a patient’s HBV infection status.
The goal is to enhance clinical decision-making and improve patient outcomes, ultimately enabling earlier care and improving the quality of life for millions of people while reducing HBV prevalence.
Preliminary findings from the ANU research revealed a high prevalence of HBV in Nigeria, with a rate of 9.5%, which is considered high according to World Health Organisation guidelines. The levels of infection also varied significantly across geopolitical zones.
To develop the machine learning tool, the researchers collaborated closely with colleagues at the Nigerian Institute of Medical Research (NIMR), which provided access to data from 916 anonymous patients.
NIMR is Nigeria’s leading medical research institute and operates a dedicated hepatitis B clinic.
The ANU research team utilized results from normal blood tests, which measure red and white blood cells, salts, enzymes, and other blood chemicals, along with hepatitis B test results.
Using the data from the 916 patients, the algorithm was able to accurately predict HBV infection with a discrimination threshold of 90%, indicating high accuracy.
The researchers then trained the algorithm to identify pathology markers that predict a patient’s HBV infection status.
The algorithm searches for common patterns among patients with HBV and applies those patterns to individuals it has not encountered before.
Once validated, the researchers aim to integrate the algorithm into routine clinical workflow as an intelligent decision support system, using a user-friendly, web-accessible app called Hep B Live Test.
The Hep B Live Test discovered that a combination of two enzymes (aspartate aminotransferase and alanine aminotransferase), patient age, and white blood cell count was the strongest predictor of HBV infection. Elevated levels of these enzymes may indicate potential liver damage.
The researchers acknowledge that the machine learning tool’s performance is currently limited to the Nigerian setting due to the data it was trained on.
They are in the process of training the algorithm with more data from other sources to assess its robustness in different settings and populations.
Early detection of HBV infections is crucial, as it improves patient prognosis and helps prevent transmission.
The current recommended test for HBV is an enzyme immunoassay, which is expensive and requires specialized facilities. In low-resource settings with limited laboratories, these tests are often inaccessible.
The ANU research team believes that machine learning could be part of the solution for public health challenges like HBV and hopes to utilize the system in the urgent fight against this vaccine-preventable disease.