Treatment Assist

Using machine learning to help mothers with HIV stay on lifesaving treatment

world icon

Context

If women living with HIV undergo antiretroviral therapy (ART) throughout their pregnancy, and while breastfeeding, the chances of passing on HIV to their child are extremely low. If these women stay on treatment throughout their lives, they are likely to remain healthy so that care for and raise their children.

In Malawi, the Ministry of Health has taken an aggressive policy stance to limit mother-to-child HIV transmission by starting all pregnant women who are HIV+ on immediate and lifelong ART, regardless of the stage of their disease.

There is growing evidence, however, that indicates that women who receive an HIV diagnosis and initiate ART on the same day are more likely to stop treatment before the end of the breastfeeding period. The data suggests that women may not be ready upon first hearing their diagnosis to commit to the lifelong treatment regimen. There are several reasons for this including HIV stigma, gender and social power dynamics, long distances to health facilities and poverty. Numerous interventions have failed to significantly raise the percentage of women who continue lifelong treatment.

microscope icon

Aim

This innovation uses routinely collected patient data to design algorithms that will allow HIV programs to accurately predict which individual are most likely to stop taking their HIV medications. This example of machine learning will enable us to proactively identify and target women who are at greatest risk of stopping their treatment and personalize their healthcare experience through enhanced HIV counseling and support. This intervention has the potential to be scaled up and applied to low-resource health settings across the Global South, with the ultimate goal of reducing morbidity and mortality from HIV, as well as the spread of new HIV infections to children.

map marker icon

Sites

Malawi (Zomba Central Hospital, Matawale Health Centre, St Lukes Mission Hospital, Pirimiti Rural Hospital, Likangala Health Centre)

map with marker on malawi
hourglass icon

Timeline

January 2018

June 2018

handshake icon

Partnerships

Funder: Pace Family Foundation

Project Lead: Dignitas International

Partners: Swansea University, CSIR Modelling and Digital Science

people icon

Innovation Team

Victor L Banda
 i2i Fellow

Athanasios Anastasiou
i2i Lead Mentor

Andreas Ziegler
Co-Investigator

Vukosi Marivate
Co-Investigator

Josh Berman
Co-Investigator

info icon

Contact

For more information, please contact Victor Banda at v.banda@dignitasinternational.org