It is expected that the algorithm, developed by the University of Southern California, will aid public health programs to better locate, and treat, people living with undiagnosed infectious diseases. The ‘undiagnosed’ aspect is important. Studies have shown how public outreach campaigns can help to prevent the spread of deadly yet treatable diseases, like tuberculosis, malaria and gonorrhea. However, these campaigns need to embrace undiagnosed patients in order to avoid further spread the disease within communities.
The new algorithm is designed to aid health policymakers in reducing the overall spread of disease by predicting undiagnosed cases. To help with cash-strapped health systems, the algorithm has been optimized to allow health agencies with limited resources, such as advertising budgets, make the best use of their resources.
The algorithm was developed by the researchers using various data, such as behavioral, demographic and epidemic disease trends, in order to generate a model of disease spread that assesses underlying population dynamics as well as contact patterns between people. The algorithm takes into account how people move, age, and die, which reflects more realistic population dynamics than existing models for disease control.
The algorithm was tested for its robustness by using computer simulations which were based on two real-world cases of tuberculosis in India and gonorrhea in the U.S.. Under both scenarios it was established that the algorithm performed better at reducing disease cases than current health outreach policies. This was primarily by sharing information about these diseases with individuals who were assessed as being most at risk.
The research was reported to the February 2018 AAAI Conference on Artificial Intelligence. The conference promotes theoretical and applied AI research and facilitates information exchanges among researchers and practitioners.