DPREDICT: An Artificial Intelligence based Prediabetes Prediction Application wins Rep. Dean Phillips’s 2021 Congressional App Challenge in Minnesota’s Third District

When asked what inspired the creation of DPREDICT: An Artificial Intelligence based Prediabetes Prediction Application, the students said, “Diabetes, a chronic disease, affected 425 million people in 2017 worldwide, with the numbers projected to increase at a rate of 7.3% per year.  According to the Center for Disease Control (CDC), 25% of the healthcare cost in the United States is for diabetes management. Each year, $237 billion is spent on direct medical costs and another $90 billion on reduced productivity. The total economic cost of diabetes rose 60% from 2007 to 2017. Early diabetes or Prediabetes is a condition where the blood sugar levels are higher than normal. Prediabetes progresses to type 2 diabetes (T2DM) in approximately 25% of subjects within 3 to 5 years, and about 70% of individuals with prediabetes will develop type 2 diabetes within their lifetime. Prediabetes has a relatively long asymptomatic phase and 50% of the cases go undetected. Early warning systems of prediabetes can allow a patient to make lifestyle changes that can slow down or prevent progression to T2DM. During the Coronavirus disease (Covid-19) outbreak, diabetics were more likely to get hospitalized or die than the general population. Another Covid-19 complication, Black Fungus or Mucormycosis, mainly affects people with diabetes. It is fatal in 50% of the cases and even if the person survives, they lose an eye or significant lung capacity. Currently, the only way to detect prediabetes is through a laboratory test every year. This solution presents many problems namely cost, logistics, availability of required tests, payment source, and last but not least the lack of general awareness in the population to get tested as this disease has a long asymptomatic phase. With the current cost of $49 per test, the cost of conducting this test worldwide will be about $392 billion/year, which is an extremely high number. Last, but not least, diabetes runs in our family. Our goal was to create a system to predict prediabetes in a person at no cost, without laboratory visit requirements, and at any time use it as many times as desired. Early detection of prediabetes would allow a person to make lifestyle changes and prevent or delay the progression of the disease.”