Over the last few years technological developments have seen medical practitioners and organizations increasingly transition from reactive to proactive healthcare. While the use of predictive models to mitigate illness outbreaks is nothing new, advances in artificial intelligence (AI) are now allowing these models to be developed into highly effective tools. Founded in 2015, AIME (Artificial Intelligence in Medical Epidemiology) is a pioneering US-based start-up using big data analytics and machine learning to predict the location and time of infectious disease outbreaks in real time. MedicalExpo e-magazine caught up with Dr. Helmi Zakariah, AIME’s Asia-Pacific CEO, to learn more about the technology and its applications.
MedicalExpo e-magazine: How does AIME work? What are its most innovative features?
Dr. Helmi Zakariah: AIME is a scalable, real-time, multi-platform system involving web applications and mobile apps, all connected through a unifying database. One of the main features of the system is a bot called REDINT, which automatically searches through more than 40 different databases for epidemiology, weather and geographical data. Processing this data using machine learning then allows the system to predict and geolocate disease outbreaks.
Where multiple outbreaks are predicted to take place simultaneously, AIME is able to create a prioritization index across the affected region, so insecticides, larvicides and human resources can be deployed most effectively. This allows public health professionals to preempt outbreaks with the best course of preventative action.
ME e-mag: How accurate is AIME?
Dr. Helmi Zakariah: For AIME to be successfully implemented it must be deployed and used throughout the healthcare industry—from community clinics through to regional hospitals. Without a continuous stream of new disease incidence data, the platform is rendered ineffective.
“During tests of our Dengue Outbreak Prediction platform in Malaysia and Brazil, the system predicted where outbreaks would occur with up to 88% accuracy, three months in advance and to within a radius of 400 meters.”
If it receives such data, however, AIME will continue to learn. Its accuracy and intervention effectiveness then becomes steadily higher. During tests of our Dengue Outbreak Prediction platform in Malaysia and Brazil, the system predicted where outbreaks would occur with up to 88% accuracy, three months in advance and to within a radius of 400 meters. This platform is now used in Malaysia, Brazil and the Philippines to help healthcare providers manage and curb outbreaks.
ME e-mag: What are your plans for AIME now? What are you working on?
Dr. Helmi Zakariah: Due to a global consensus on disease surveillance and initiatives such as the World Health Organisation’s International Health Regulations, huge quantities of health and disease-related data are already being generated. However, the skills and capacity to analyze such data are not as developed.
Using AI to discover patterns and markers hidden from plain sight and conventional statistical analysis has opened up an array of exciting new possibilities. One example is a new project of ours focusing on the prediction of antibiotic resistance. This represents a huge departure from the conventional diagnostic approach to combatting antimicrobial resistance. We’re betting that AI can become just as diagnostic as a rapid test kit, but instead of using blood we use data to gauge probability. We also want to create devices to diagnose tuberculosis and malaria, and software to diagnose diabetic retinopathy.
“We’re betting that AI can become just as diagnostic as a rapid test kit, but instead of using blood we use data to gauge probability.”
ME e-mag: Will there come a day when we can predict all disease outbreaks before they occur?
Dr. Helmi Zakariah: Technology is now developing at an exponential rate. Just over a decade ago contracting HIV was a death sentence – now one is more likely to die from a car accident than from AIDS (maybe we’ll see this statistic change as autonomous driving evolves). If we have the right sources of data, the right regulatory framework and the right approach to data exchange, I don’t see why we can’t predict cholera and other disease outbreaks in the same way that we can currently predict a tsunami occurrence.