We are an Software company in the healthcare space and in our experience we noticed a lot of gaps in the health care industry. The main gap being insufficient doctors and medical specialists, especially in Africa. In Kenya for instance, the ratio of doctors to patients is 1 to 16,000 while the recommended ratio is 1 to 300 patients. With technology we saw an opportunity to bridge this gap by aiding diagnosis for both patients and doctors/medical specialits .
What it does
Uzima Health is a Big Data & Machine Learning powered solution that helps patients and medical professionals in assisted diagnosis and maps patients to health facilities and health workers based on their medical conditions and regions. A Patient simply dials the Uzima Health shortcode or accesses via the App, Uzima Health gets the patients location, the patient is prompted to enter how they are feeling or choose from a predefined list. Uzima Health attempts to map a patient to a nearby facility based on the patient symptoms and also the nearest facility. A patient can enter their symptoms in their native language and using Machine Learning Uzima Health is able to translate the patient's symptoms into English, from this translation the system is able to perform a look-up on the possible medical condition a patient is suffering from. Incase a health care facility is too far, for instance in remote areas, Uzima Health links the patient with a registered local community health worker who is nearest them. These local community health workers can guide the patient on what to currently do as the patient waits to be taken for a medical facility, and these small steps can save lives. With the data collected through Uzima, Uzima is able to identify areas with outbreak of certain diseases e.g The Corona Virus, special conditions affecting specific regions and share this data with relevant parties. e.g Governments, Health Agencies, NGO'S etc.
How we built it
We first started by building a mobile application to help in remote monitoring of patients with chronic illnesses e.g. diabetes. When we deployed this solution it made a huge impact on our first client, which is a hospital in Nairobi. It improved monitoring of patients, reduced operation costs, improved efficiency as well as patient and doctors interaction. From these results we felt we could do more in the healthcare sector and that is how Uzima Health was born. We incorporated new features based the experience from the initial version. These features entailed:
- Introduction of Machine Learning Introduction of Machine learning to help in detection of potential ailments from patient’s symptoms and their locations. A patient can enter their symptoms in their native language and using Machine Learning Uzima Health is able to translate the patient's symptoms into English, from this translation the system is able to perform a look-up on the possible medical condition a patient is suffering from.
- Data Analytics Upon collection of this data it is used for analytics such as: Detection of Diseases outbreaks, Prevalence of diseases and existence of special conditions in certain regions. As an example this can be and would have been used to trace outbreaks of viral diseases e.g. COVID-19 and respond in the very initial stages of such diseases or any other that my arise.
- Sharing of Collected Analytical Data Data Collected such as possible outbreaks, number of new and existing infections, number of attended and unattended cases, available health care providers/workers/specialists in a region and the patient to doctor ratio can be shared with relevant stakeholders for purposes of planning and mitigation. ## Challenges we ran into
- Difficulty in getting access to hospitals, clinics and healthcare providers to pilot our solution in the initial stages.
- Inadequate resources; Human & Financial Capital to fully implement and actualize the envisaged Uzima Health Solution.
- Difficulty in acquisition of Data sets to train our models for native language translation and diseases diagnosis. ## Accomplishments that we're proud of
- Over 500 patients who are currently being monitored remotely by their doctors from 3 hospitals in Nairobi and 10 clinics. These hospitals and clinics are able to share medical records for patients among each other hence ensuring accuracy of patients data, faster and more efficient patients diagnosis.
- We have built the Native Language translation module and are currently training more models to understand various local local languages, currently in Kenya.
- We have built the engine that is capable of getting possible diagnosis from symptoms provided by patients and health care workers.
- Our engine is able to map patients to nearest health care centers based on their diagnosis, location and services provided by the health care centers. If a health care center is not nearby, the engine is able to connect a patient to the nearest community health worker who can in turn provide first aid and guidance. These community health workers are registered with healthcare providers and thus can request for further assistance if need be e.g. request for an ambulance.
What we learned
- The need to partner with healthcare stakeholders such as hospitals, healthcare workers and medical research institutions. e.g. Medical research institutes proved to be vital in provision of data sets to aid in training of the models for disease diagnosis.
- There are few health care practitioners compared to the number of patients who need to access their services. This practitioners are also not evenly distributed further making it difficult for patients to access their services. With leveraging technology, our solution is able to bridge this gap. ## What's next for Uzima Health With the data collected through Uzima, it will be possible to identify areas with outbreak of certain diseases, special conditions affecting specific regions and share this data with relevant stakeholders. 1.Fully open sourcing the solution to incorporate more collaborators and thus allow faster development,incorporate new ideas and technologies from different parts of the world. 2.Adding more data to the disease diagnosis module and performing more training/testing of models. 3.Leveraging Machine Learning and Data analytics will make the process of analyzing huge amounts of data from different regions faster, easier, efficient and more accurate. 4.Partnering with more healthcare facilities, research institutions to improve the disease diagnosis models.
- Rolling out of the project on a global scale.