Medical AI Record System (MAIRS)

As of the mid-2023 , there were 110 million people forcibly displaced worldwide according to the UN Refugee Agency (UNHCR).  The African continent includes 18 million refugees and 44 million internally displaced people (UNHCR/Africa)

One of the challenges facing refugees and displaced people  is access to   primary health care timely.  This creates a double burden: long wait times due to limited resources leaving patients struggling to receive timely medical attention, while healthcare staff become overburdened managing patient flow and maintaining medical records.

We provide an AI-driven solution to assist healthcare staff archiving medical records. 

The solution converts patient-doctor interactions from audio input  into textual medical records.

Imagine a ten-year-old refugee laying on an examination table in a refugee camp and mumbling something in her native language.  A translator sitting beside her, translated to an NGO doctor: "She says her stomach hurts again." The child had been complaining of stomach aches for weeks, and a lack of medical notes from her previous consultations made diagnosis a challenge. 

The proposed solution automatically transcribes the vocal  interactions between the doctor, the translator, and the child  into textual medical records. At the beginning of the consultation, the doctor pushes a button to start audio recording through a smartphone. The audio file will be processed by an AI model that transcribes audio into text, generating medical notes according to a specific template ready for archiving.  The doctor can review, edit, and approve the medical notes content. 

Technically, we refined our AI model on audio medical interactions in order to accurately transcribe audio files into textual medical notes.

The target population is  refugees and internally displaced people (IDPs) residing in camps and settlements across Africa. Many of these individuals have fled ongoing conflicts in countries like Central African Republic, Nigeria, South Sudan, and Burundi. Their large numbers within the camps create a significant burden on healthcare staff, making it difficult to deliver timely and appropriate medical care.

By automating medical note generation, we can streamline the medical documentation process, freeing up valuable time for healthcare staff. This translates to a lighter workload for healthcare staff, allowing them to see more patients per day. As a result, patient waiting times can be significantly reduced, leading to faster access to primary medical care.

We are a unique team united by a shared mission: improving healthcare access for refugees and displaced people in Africa. Our strength lies in our diversity, bringing together expertise across technology, medicine, and management.

Our technical team is composed of  African AI Experts.  Our team of researchers are at the forefront of artificial intelligence (AI) research, specializing in solving Natural Language Processing (NLP) challenges specific to African languages. We are Committed to Africa's Growth. We are actively involved in strengthening the African machine learning community, serving on boards of prominent grassroots movements like Masakhane and Deep Learning Indaba.

Our Medical advisor is a Family Physician. We are fortunate to have a board-certified family physician on our team, providing invaluable medical insight and ensuring our solutions are grounded in real-world healthcare needs. We also benefit from the guidance and experience of seasoned NGO staff who have extensive experience working directly in refugee camps across Africa. All team members are native Africans, having grown up and lived close to the refugee and displaced people context we aim to address. Together, we create a bridge between cutting-edge technology and the urgent needs of Africa's displaced populations. Our diverse backgrounds and shared passion fuel our commitment to making a positive impact on people's lives.

To ensure our solution truly meets the needs of African refugee camps, we're launching a pilot project in Kenya.  This controlled rollout with a smaller group of NGOs healthcare staff will allow us to gather valuable feedback and refine our automatic medical note generation system. By working closely with healthcare staff on the ground, we can optimize the system for real-world scenarios and maximize its effectiveness in improving healthcare delivery for refugees. 

Despite the hurdle of limited investment, FairConnect has made significant strides in developing and validating its solution:

• Multilingual Language Model: We've successfully built a core component – a multilingual language model capable of handling the diverse linguistic landscape of African refugee camps.
• Streamlined Transcription Pipeline: A robust transcription pipeline has been established, ensuring accurate conversion of vocal consultations into written text.
• User-Friendly Web Platform: We've developed a user-friendly web platform to facilitate testing and future implementation of the automatic medical note generation system.

FairConnect isn't just about theoretical innovation. We're actively testing our solution in a real-world setting.  Our pilot project is currently underway in Kenya, where doctors are using the system in both English and Swahili. This pilot will provide crucial feedback that will enable us to further refine the system and ensure it seamlessly integrates into the workflow of healthcare staff across Africa. The success of the Kenyan pilot will pave the way for wider implementation, ultimately improving healthcare access and reducing waiting times for refugees in Africa.

We have reached a juncture where it's crucial to enhance our capital, network, and resources in order to facilitate development and deployment.

One major challenge we face is securing partnerships with NGOs and medical collaborators. The Solve network will facilitate the establishment of new connections with managers of NGOs operating in refugee camps and settlements across various African countries, enabling us to initiate pilot projects. Additionally, we seek input from healthcare professionals with extensive experience working in refugee camps.

We also require partnership and mentorship from Solver  teams who have successfully built healthcare platforms and information systems. Leveraging their expertise will help us ensure that our solution seamlessly integrates with the systems they have developed. Furthermore, we seek guidance from legal experts to ensure that our solution navigates the healthcare landscape effectively. Collaborating with experts in these fields will enable us to ensure that our solution meets the ethical and legal requirements of healthcare data management.

AI-driven medical notes transcription is taking off, but there's a bias towards wealthy nations. This means patients who don't speak these dominant languages face challenges. Consequently, African countries, particularly for refugees and displaced individuals, are being overlooked in this advancement.

The majority of  refugees and displaced individuals today have to wait for a long time to have access to primary healthcare. With efficient implementation of AI-driven medical notes transcription could dramatically cut down on how long refugees and displaced individuals wait for healthcare.

The MAIRS solution focuses on the gap between AI-driven medical notes transcription solutions and African refugees and displaced individuals. MAIRS is empowered by AI models refined for African languages to efficiently implement  automatic medical notes transcription. Once integrated in the refugee healthcare process, MAIRS will slash wait times significantly.

The success of integrating our AI solution in the healthcare system for refugees will foster the adoption of AI solutions in healthcare applications in the African continent, and potentially other non wealthy countries.

FairConnect uses digital tools and automation to improve how health data is collected and managed in refugee camps and settlements in Africa. This approach is expected to make healthcare delivery more efficient and effective for staff, ultimately leading to better health outcomes for refugees.

FairConnect's core strategy is to move away from relying solely on spoken medical information to a system using integrated digital notes. This switch within NGOs and public health facilities makes record-keeping and transcription much faster. By eliminating manual data entry and paperwork, FairConnect allows for easier tracking and real-time access to accurate, up-to-date medical records for refugees.

FairConnect tracks its success by focusing on two key UN Sustainable Development Goals (SDGs) related to health:

• SDG 3.8.1: This goal aims to increase access to essential healthcare services. FairConnect helps achieve this by making it easier for doctors in refugee camps to deliver care.
• SDG 3.d.1: This goal focuses on strengthening how countries manage health risks. FairConnect's system allows for quicker identification and response to potential health issues in refugee camps.

Our approach also supports SDG 9, which targets advancements in technology and collaboration. FairConnect leverages technology to improve healthcare delivery in developing countries, fostering global partnerships in the process.

To track how effectively FairConnect improves healthcare and supports the SDGs, we continuously monitor and assess data.  We look at key quantitative metrics like the number of consultations happening in refugee camps, how long each consultation takes on average, and the number of medical records created. This data helps us understand how widely FairConnect is being used and how well it's  supporting doctors in delivering care.

We'll also be gathering feedback from healthcare workers about how easy and effective they find FairConnect to use.  This will involve surveys, interviews, and focus groups. These regular evaluations will help us understand what's working well with FairConnect, where it's making a difference, and where we can improve things even further.

MAIRS core technology lies in automatic audio transcription technology. This innovative approach eliminates the need for manual note-taking by doctors in refugee camps:

• Machine Learning: FairConnect builds machine learning models trained on massive datasets of spoken language.
• Training for Accuracy: The model is constantly learning and improving its ability to recognize sounds, words, and even sentence structures within conversations.
• From Speech to Text: By analyzing the audio recordings of medical consultations, the model can generate accurate written transcripts, capturing the details of each patient-doctor interaction.

This technology empowers healthcare workers to focus on delivering quality care,  rather than being bogged down by paperwork.  The resulting transcripts streamline medical notes and provide a clear picture of a patient's medical history, ultimately leading to better healthcare outcomes for refugees.

full-time staff: 3

part-time staff: 1

other workers: seasoned NGO staff

FairConnect has been working on this solution since February 2023. Nevertheless, the technical team has been working on the core technology for more than 4 years.

FairConnect is built on a foundation of diversity. Our leadership team reflects a range of genders, ages, thinking styles, and cultural backgrounds. This allows us to see challenges from multiple perspectives

The solution was created using the information gathered from healthcare professionals and refugee communities. Community health professionals who come from a variety of target communities make up the collaborators at our startup.