Artificial Intelligence (AI) chatbot to improve disease education of People with Haemophilia in Sub-Saharan Africa

Haemophilia  is a congenital bleeding disorder affecting 1 out of 10,000 people in the world. Given the fact that the disease is rare, this induces many challenges for People with Haemophilia, the mothers (obligate carriers) and their families.

In the case of Sub-Saharan Africa (SSA), Haemophilia care needs to be significantly improved and many barriers exist including a lack of diagnosis, poor disease knowledge, and high stigmatization. In addition, the influence of cultural beliefs and practices trigger behaviours such as social exclusion or inadequate care.

Serious gaps in knowledge were observed and defined as the main  cause of the other several key challenges. There is an unanimously  rgent need to improve education, particularly regarding the inheritance pattern and risks management of haemophilia at home in Senegal for example. As known, Patient education (PE) should be the overarching goal to ensure better hemophilia care and management in SSA as per the guidelines of the World Federation of Haemophilia which in its Principle 4: (Education and training in haemophilia) emphasized the importance of integrating effective disease education for patients. 

Consequently, this initiative aimed to tackle the gaps in education in SSA through the design and implementation of an original educative AI- chatbot in three languages ( French, Wolof and English).  We believed that by improving the education, this will also be a catalyst to empower PWH and generate positive outcomes on their disease management and the overall QoL. 

This project encompasses various dimensions that can be summarized as follow:

1. The first key step consisted of a research phase and field study which helped to understand the main barriers in access to care for People With Haemophilia in Senegal (2018-19). Results have been published and can be found here: https://onlinelibrary.wiley.com/doi/abs/10.1111/hae.14040
2. The crucial need to improve education amongst the haemophilia community in Senegal led to the ideation and implementation process of a concrete solution that embraces the potential of digital health and Artificial Intelligence. A sufficient level of digital literacy and access to smartphones of the main targeted populations were positive indicators to start the initiative.
3. A patient-led approach was undertaken to design and implement an AI chatbot with a first pilot in Senegal.  A task force was put in place including  People with Haemophilia (PWH) of the Senegalese Patient Association,  family members and especially the mothers also called obligate carriers. The task force was involved in all the steps of the initiative from the content creation to the effective implementation in Senegal.
4. The first primary outcome is the successful creation of an original AI chatbot in French and Wolof named Saytù Hemophilie incorporated in a mobile app and available for Android and Apple iOS devices. The official launch was done during the World Haemophilia Day 2022.
5. Technical development : Different latest Artificial Intelligence subsets and models, such as a chatbot builder using Natural Language Understanding (NLU), Automated Speech Recognition (ASR) technology, Wav2vec model React Native platform, were used to produce the first validated prototype of a specific chatbot.The fact that there is no specific chatbot in Wolof led to the creation of a unique framework in this case.
6. Features: Saytù Hemophilie has the ability to understand and respond in Wolof and French  of questions related to treatments, transmission, symptoms management etc. taking into account a large database of intents. Saytù Hemophilie can support PWH in case of emergency by providing information about the closest center with available clotting factors and by directly connecting the patient to a medical doctor. Conversations with the AI chatbot are stored for the Wolof version, allowing users to come back at anytime to improve their knowledge. PWH and their families can also use the platform to disseminate information within their families.
7. Security: Saytù Hemophilie does not collect any health data, this avoids the common issues related to data privacy.
8. Trainings of PWH and families, medical doctors of the Blood Transfusion Center were performed in Senegal during focus groups in the medical centers and during home-visits and communication on social media
9. Evaluation : A first evaluation of the usability and satisfaction was performed. Saytù Hemophilie was assessed as highly usable with an average SUS (System Usability Scale (SUS) score of 81.7, above the benchmark for average usability of Digital Health Apps (DHAs). Full data is available here: https://onlinelibrary.wiley.com/share/6R9E89PNZZ24QJUEFEJF?target=10.1111/hae.14815. The second stage includes an assessment of the impact of the chatbot on the level of education. First results should be available in Q2 2024.

Perspectives 

-Expansion in other countries: The solution is in the process of scaling up to extend support to more People with Haemophilia in additional countries. Considering similar gaps identified in many African countries, Ivory Coast, South Africa,Kenya, and the Gambia have been selected for the first stage of the initiative.  The chatbot is now available on Ivory-Coast, Gambia and Kenya though the full dissemination campaigns and activities will only really start in Q1 2024. South-Africa should be launched in Q1 2024.

Other identified steps needs for which additional  funding would be required, as examples :

–Leverage on AI capacity capabilities and LLM to improve the performance of the chatbot while ensuring safety and accuracy of information. Add new features  and more languages to improve the user’s experience and ensure interoperability with health infrastructure of medical treatment centers.

-Additional  local trainings in various cities in each country to reach a wider population.

– Integrate additional rare blood disorders—leukopenia, von Willebrand Disease (VWD), thalassemia and sickle-cell disease (SCD) for patients living in low-resource settings, with a particular focus on the African context

Sustainability: Ensure to work closely with Ministries of Health to foster national adoption of the initiative, encourage local ownership by the PWH and medical doctors in the targeted countries as examples of key strategies foreseen so far though more approaches are being explored.

The main added values of the solution can be summarized as below:

• Saytù Hemophilie, is the first AI chatbot for haemophilia developped in a local language of a SSA country namely i.e in Wolof.  Languages can be a barrier to the adoption of digital tool  as known and culturally adapted solutions have been proven to enhance better compliance, adherence but also motivation to use the solution. The first results of usability and satisfaction highlighted that Wolof version was the most appreciated vs the French as expressed to be closed to the PWH and carriers’ needs.
• In this specific case, Saytù Hemophilie has the ability to understand and respond in Wolof  taking into account a large database of intents based on the population’ expressed needs. This also enabled the chatbot to understand the questions from the user in various way and the chatbot was built to reflect the disparity of spoken Wolof observed within the population.
• Two other languages are also available French and English available for both Android and IOs devices and the aim is to ensure to integrate as much as possible local languages in the countries where the initiative will be launched.
• The fact that there is no specific chatbot in Wolof led to the creation of a unique framework, as already described above.
• This is an interactive and self-learning tool which provides comprehensive educational information about haemophilia, first-line care, and self-symptoms management at any time
• The chatbots’ versatility, usability, responsiveness, and ability to mimic human behaviours that could support a safe learning environment were identified as key advantages, leading to our choice of this particular digital health intervention in Senegal as a start.
• The approach of developing a specific chatbot with verified sources of information, unlike a generic one, ensures data accuracy and the safety of the users.
• First patient-led AI digital solution for Haemophilia in low-resource setting.

NB: The numbers of beneficiaries cited below are linked to the number of downloads of the App to date