Machine Learning Can Help Win the Worldwide Battle of Diabetes Detection

Diabetes machine learning

Left unchecked, the diabetes epidemic will continue to be a leading burden on healthcare systems. More than just cost, overstretched clinical resources, and the impact of absenteeism in the global workforce, some estimates indicate that at least $825 billion is spent globally trying to combat this disease.1 Tragically, diabetes is responsible for approximately five million annual deaths worldwide (approximately one diabetic dies every six seconds).2


In March 2019 a study published by PLOS Medicine concluded that those at greatest risk for diabetes are from low and middle income countries (LMICs), where the prevalence of the disease is rapidly increasing.3


The study examined the state of diabetic care in 28 LMICs, where diabetes carries 75% of the global burden. An examination of 800,000 adults in these countries revealed a failure to detect diabetes in 4.8% of the sample population, representing over half of the sample’s total diabetics. In these LMIC countries, only 1 in 4 people with the disease will get detected and have the support and care to maintain adequate glycemic control.


Diabetes presents as great a concern in upper-middle income countries as it does in LMICs. Approximately one-third of the U.S. population is classified as having prediabetes, and over 1.7 million of those are likely to develop diabetes this year.4 Diabetes is increasingly a global concern. And key to combating its spread – and the added burden of downstream complications – is to prevent the onset of the disease in the first place.


This is where machine learning can prove invaluable in the clinical environment. By applying predictive analytics tools to EHR and other ordinary clinical data, healthcare providers can flag patients with prediabetes who are at high risk for developing diabetes within an established time frame. This creates intervention opportunities where preventative measures can potentially delay or even prevent the onset of type 2 diabetes.


Only a highly specialized, IT-based clinical predictor — with validated modeling — can analyze vast amounts of data to identify and stratify prediabetic populations at highest risk of progressing to full diabetes within a specific time frame. Singling out a small subset of individuals, Medial EarlySign’s prediabetes to diabetes (Pre2D) AlgoMarker™ offers healthcare organizations the ability to prioritize patients at high risk and allocate resources accordingly where intervention could be most beneficial.


Directing resources where they are most likely to be effective can result in better care delivery and vastly improved patient outcomes. Regardless of the setting or location, diabetes progression could be delayed or halted, potentially saving millions of lives and billions of dollars.


__________________________
1Worldwide trends in diabetes since 1980: a pooled analysis of 751 population-based studies with 4·4 million participants,
The Lancet, Vol. 387, Iss. 10027, pp.1513-1530, Apr. 9, 2016 DOI:10.1016/S0140-6736(16)00618-8
2International Diabetes Federation, What is Diabetes? Facts & Figures (2017)
3Manne-Goehler J, Geldsetzer P, Agoudavi K, Andall-Brereton G, Aryal KK, et al. (2019) Health system performance for people with diabetes in 28 low- and middle-income countries: A cross-sectional study of nationally representative surveys. PLOS Medicine 16(3): e1002751. DOI: 10.1371/journal.pmed.1002751
4Yudkin, JS. Prediabetes: Are There Problems With This Label? Yes, the Label Creates Further Problems! Diabetes Care 2016 Aug;39(8):1468-71. DOI: 10.2337/dc15-2113

Share:

Share on twitter
Share on linkedin
Share on facebook
Share on email

More like this

How Healthcare AI/MI can Augment Care Delivery

Are you looking to solve high-burden disease challenges?