Elisee JAFSIA

Discussion between the Artificial Intelligence panel's speakers.

Other people: Yo Yehudi Gemma Turon

Typhoid fever is one of the severe infectious human diseases in Africa. Out of an estimated 11–21 million cases of typhoid fever and 200,000 deaths occur worldwide each year. Other elements contribute to this situation and particularly in rural areas where patients to doctors ratio is very low, lack of medical facilities and costly tests. There are a number of tests available presently, from molecular to immunological and biochemical to microbiological. However, Users are unsatisfied due to delays in getting test results and Imprecise diagnosis. Misdiagnosis is usually experienced since most health care facilities use only Widal test without confirmation of results with a second test method. In addition, the diagnosis of Typhoid involves several levels of uncertainties. Patients cannot tell exactly how they feel, doctors and nurses cannot tell exactly what they observe. There is therefore, an urgent need to develop a rapid, highly sensitive and cheap diagnostic tool for diagnosis of typhoid fever.

The absence of a reliable diagnosis pushes rural populations towards self-medication with all the consequences that this entails, in particular drug poisoning which can lead to death. As such, we want to develop a new method to diagnose Typhoid early, quickly, and with accurate results at a low cost. We propose a hybrid diagnosis method with a Deep Learning algorithm applied on blood serum images which seems to be the best accepted test by laboratory technicians. We use CNNs as algorithms on the collected images to train the algorithm. We use a second algorithm Rough Set Theory. This algorithm based on symptoms variables which are structured data collected from medical doctors allow us not only to confirm the diagnosis but above all to determine the level of severity of the disease and serve as a Decision Support System. The two algorithms will be part of a mobile app connected to an optical microscope, the phone camera will scan images from the slides on the microscope and classify them into Normal or Infected. In a series of rapid spotlights, we present tools and frameworks that serve as cornerstones for the envisioned diagnostics system including: 1) Values, influences and forces that shape the diagnostics 2) AI solutions to be used; 3) Data sources 4) Ethics and governance model.

The lack of accessible quality healthcare is one of the biggest problems in Africa and other developing countries. This is not only due to the unavailability of resources, but also to the absence of a structured formative process for the design and management of healthcare facilities. This situation strongly contributes to deepening inequalities in access to quality healthcare. Through an Open Society Foundations funded project, the Mboalab aims to remedy these inequalities by building Open-Source devices for medical labs. Local manufacturing can address the infrastructural barriers that prevent imported or donated equipment from being properly used, and can facilitate the diffusion of innovation into healthcare practice. This project is part of the large MboaLab mission to contribute to the Common good and catalyze sustainable local development through Open Science.

Open science is the best and fair approach to support local manufacturing. That is why the crux of our approach is the use of “open source hardware”, where designs for easily replicated, high quality diagnostic tools are shared with the potential to transform medical devices through the use of digital fabrication and inexpensive, well-engineered parts from mass-produced consumer goods. During FOSDEM, we will present the different facets of our project: 1) the set of prototypes of high quality and inexpensive open-source devices we are going to build; 2) the capacity and capability building dimension of the project, enabling the empowerment of young cameroonians; 3) Expectations of the project.