Authored by: Laura Kanov, Senior Vice President, Product Strategy

My good friend lost his leg to diabetes almost two years ago. He has been going through physical and occupational therapy for months to regain mobility with his prosthetic leg. He has finally graduated to forearm crutches. He has been adjusting and struggling to do normal things like walk, get in the car, drive and play with his son. He has fallen in public and relied on strangers. His marriage fell apart.

Another friend with diabetes did not want to be a burden to his family, refused amputation and suffered a painful death from his gangrenous infection.

Diabetes is an insidious and deadly condition. In 2015 [1], 30.3 million Americans, or 9.4% of the population had diabetes with only 1.25 million of them considered type 1. Of the 30.3 million adults with diabetes, 7.2 million were undiagnosed. An additional 84.1 million Americans aged 18 and older had prediabetes. It is important to realize this all year, but especially today in honor of World Diabetes Day.

My mother developed Type 2 Diabetes, but was very, very good about watching her diet and taking her meds. Still, she suffered a detached retina late in her life. I am my mother’s clone. Despite my normal BMI and intentional sugar avoidance, my blood sugar has been on the rise, so I have been investigating ways to reduce it.

Along the way, I’ve learned a lot.

  • There is sugar in EVERYTHING! (it seems). Read your labels. Packaged and processed goods are notorious for using sugars. Stick to fresh foods as much as possible. Carbohydrates, when metabolized turn to sugar, so watch your total carbs, too.
  • It’s not just BMI, it’s muscle mass that will help reduce your blood sugar [2]. Skeletal muscle uses glucose in the blood to produce energy. For every 10% increase in your skeletal muscle index, there is an 11% reduction in insulin resistance and a 12% reduction in the risk for prediabetes.
  • Even if your blood sugar levels are normal, your pancreas could be working overtime to produce enough insulin to control increased sugar and/or overcome insulin resistance. At some point this is not enough and blood sugar levels become elevated but stable. But as the disease progresses, an individual experiences instability and eventually decompensation [3].

Perhaps the most important thing I’ve learned, is that diabetes is controllable and the early stages are reversible. Do not take this disease for granted. Please educate yourself, your friends and family.

Here at HBI, we’ve also done an immense amount of research into the subject. HBI creates custom, machine-learned predictive models for our clients that include the risk of Type 2 Diabetes in the future 12 months. This algorithm applies a risk score calibrated to the probability the individual will develop the condition (as evidenced by a diagnosis). We’ve found that individuals with high probabilities often have the disease, but have not yet been diagnosed, alerting caregivers to reach out for confirmation and intervention.

HBI’s custom models have found individuals with elevated blood sugars to be 5-7 times more likely to develop a Type 2 Diabetes diagnosis in the next 12 months, and individuals with a BMI of 36+ to be 3 times more likely.

Other model insights show interesting significant influences on a diabetes risk, including anemia, atherosclerosis, diseases of the veins and heart disease.

1. NLP-based trained algorithms can be used to identify more diabetics than code-based identification alone [4].

In other words, patients with diabetes don’t always have a confirmative diagnosis code in their medical record, but diabetes-related clinical terms, medications or lab results are often hidden in the physician notes and observations. Our Natural Language Processing-based case finding algorithm used these notes and observations to further identify diabetic cases. In fact, using this algorithm in the study we found 9% more patients with diabetes than using diagnosis codes alone.

2. Critical transition states can be identified in EMR data (much like they can be found physiologically) [5].

In other words, when HBI analyzed patients with a confirmed diagnosis of Type 2 Diabetes, and their pre-disease clinical history, we were able to identify a “signal” in the EHR data six months prior to the first confirmative diagnosis. That means it is possible to alert providers to a window of opportunity during which they can apply proactive interventions to delay or even prevent the onset of Type 2 diabetes.

Millions of people live long, healthy and happy lives with diabetes around the globe. Whether you are one of them or not, I hope you take this time to reflect on how important it is to engage in a healthy lifestyle for yourself and for those who care about you. Visit for more information.