CARE Tech Forecasts Diseases You’re Likely To Catch

Hypochondriacs rejoice! The software uses medical history, records, and experience to discover whether you’re at risk for as many as 100 diseases. But will you follow your doctor’s advice to prevent them?

girls with medical masks on


What if a software program could give you a list of the top 20 or even 100 diseases that you were likely to develop in your lifetime? Imagine walking out of a doctor’s office with a complete disease risk assessment profile along with a wellness plan to nip all of them in the bud.

Professor Nitesh Chawla at the University of Notre Dame hopes to offer such an assessment with his software engine called CARE (Collaborative Assessment and Recommendation Engine)–proactive medical care that emphasizes disease prediction and preventive treatment, well before diseases manifest.

The practice of medicine is mostly reactive and disease driven, treating illnesses after they have emerged, and doctor’s risk assessments are mostly sporadic educated guesses based on experience, memories, physical examinations, and the patient’s family medical records. Chawla wants to change how prospective health care looks, not only giving doctors an effective tool to help them take care of their patients, but also empowering patients to own their medical futures with the knowledge of diseases they are susceptible to.

“The goal is to determine and minimize an individual’s risk,” says Chawla. “We want to provide every patient with a personalized answer to the question: What are my disease risks?”

CARE uses the vast amounts of data already present in the health care system in the form of alphanumeric designations called ICD-9-CM codes, which are attached to each condition a person is diagnosed with. The program compares codes in a patient’s history plus current symptoms and any lab results with other patients with similar disease profiles to draw up a list of the top 20 or 100 most likely diseases, along with probability factors of how likely they are to contract it.

“Lets say that a person X is diagnosed with eczema, hay fever, hypertension, and diabetes,” says Chawla. “The system prefers a minimum of three diseases to give more matching. The CARE/ICARE system finds all patients similar to X that share those or a subset of diseases in common with person X. Then, it looks for what other diseases those patients have developed.”


The premise here is that diseases just don’t develop in isolation, but have their roots in a person’s lifestyle, genetic risk factors, and their environment. Using the medical histories and experiences of patients with similar initial diseases, the CARE engine outputs a ranked list of future disease risks; the ranking depends on how similar the diagnoses of the others patients are in comparison to the patient in consideration.

Expert critics, however, say ICD codes don’t reveal any data about physical examinations, environment, and medical history. “In essence the important cues and clues to diagnosis are not found in ICD codes,” says Dr Greg Latchaw, an obstetrics and gynecology specialist in South Florida. “Moreover, ICD coding which is assigned sometimes by a physician but oftentimes by a non-physician clerk is notoriously subject to inaccuracies.”

Chawla admits that ICD codes have issues, which are well documented, but says that they capture the actual documented diagnoses. He is currently incorporating additional data such as notes about physical examinations, lab results, and medical history, to complement and supplement the ICD-9-CM codes.
Tom Liddell, CEO of Michiana Health Information Network who is collaborating with Professor Chawla, believes that the CARE program could be the ”beginning of a network of networks” where multiple data sources and analytics could go steps further in helping their patients. So long as patients actually do what they’re told to prevent likely disease.

“The larger issue is not whether they’re reliable or not, more so whether they are effective in getting patients to comply with lifestyle changes,” says Dr Richard V. Lee, Professor, Department of Social and Preventive Medicine, University at Buffalo. “A majority of patients I see under the age of 50 feel as if they’re bulletproof, and even with hard evidence demonstrating they are at risk, most still avoid making changes recommended by their physician.”

Chawla says that he doesn’t realistically expect all patients to sign on. But when those who do start seeing how their actions are affecting their health, they will gain more confidence in the system and accept the doctor’s recommendations. “Seven out of 10 deaths among Americans each year are from chronic diseases,” says Chawla. “Even if a relatively smaller proportion of younger patients appreciate the impact and recommendations, it can lead to a significant impact in terms of wellness, as well as time and resources.”

[Image: Flickr user mikeleeorg]


About the author

Freelance Science and Technology Journalist