Kioni bot rocked the hall as an optimal solution to collect missing sociomedical data to fight HIV drug resistance
Let’s put ourselves in context by telling the story of Adama. She is a Kenyan who went to a funeral that ended up becoming an orgy, as it is tradition in Homa Bay. Her friend Bria asks her: “have you done the HIV test?” To do so, she would have had to walk 5 hours to a health facility and then wait 7 hours to be treated.
Healthcare friction is one of the causes that lead patients to neglect their rigorous daily HIV treatment. In consequence, the drugs that help to deal with this life-long illness become ineffective, leading to the emergence of HIV mutations and HIV drug resistance. (Short and beautiful video: HIV drug resistance explained).
Driven by the need to prevent a new crisis in the HIV epidemic, Johnson & Johnson organized the HIVhack, a series of workshops that concluded with a 48 hours hackathon. The data sets focused on Africa, where the illness is a threat. The events were hosted at DigitYser, the clubhouse of the tech communities of Brussels, where I have the pleasure of been taking part of its data science bootcamp.
Modeling day-to-day friction healthcare in Africa
Kioni bot is a messenger interface that informs patients where to find health facilities that match the treatment they are looking for. To avoid the issue of HIV stigma, the service is also provided for different illnesses (e.g. malaria, dengue). The criteria to define the optimal health location was based on the treatment, the travelling distance and how crowded a facility is. By using the database from D.R. of Congo (kindly provided by Bluesquare), Kioni bot could recommend locations of human settlements that in some occasions are not even registered in google maps. Healthcare data from Africa contain rich descriptive analyses. However, these data are hard or impossible to model because key information is missing.
We need a change of perspective from governments,
world health organizations and NGOs toward data
collection and use. Its main purpose has to be data
science analysis, not descriptive reporting.
Bots are a friendly and cost-efficient solution to gather data. They mimic a conversation with the user and by collecting their replies, they create a new database. Once connected, bots and users can keep interacting. Kioni bot offers a win-win relation where patients can receive psychological support, have info about stock-outs and more. For the stakeholders, Kioni can provide new and unique relevant data to model HIV drug resistance.
This hackathon was a lucky opportunity to team up with Laurens van der Cruyssen (psychologist) and Lantana, a startup that helps businesses to transition toward AI and blockchain. Nothin' but a dream team. Two other teams were also nominated for the second phase. Finally, it was a pleasant experience to scream "Let's empower Africa" on the mic after we were announced as winners.
The prize is an invitation to visit, together with two other teams (WorldWise Wild Hack and ARTVEC), the offices of J&J and present a subsidy proposal to develop the project. Cheers and congratulations to all the DigitYser bootcampers who dedicated their time to this tricky project!
Update: After visiting J&J and present our subsidy proposals, J&J did not finance any of the proposed projects. After some interviews with possible stakeholders for Kioni bot, it was understood that the pharma industry is not the best friend of NGOs or hospitals. However, it was a nice research experience.