AI Think Biosolution

Product/Service – Our Chronic Disease Prevention Platform is CE-Marked, FDA CDSS, HIPAA and GDPR compliant.

Role of AI in the product – Our Chronic Disease Prevention Platform has three AI-powered risk indicators HyperDetect, CardiacDetect and DiabetesDetect.

How does the AI-powered risk indicators work with current clinical workflows?
Long term facilities with built-in wellness programmes require nurse managers to monitor patient vitals on a weekly basis to check for transitions to critical stages, in order to avoid patient hospitalisation. Long term facilities also collect patient vitals (blood pressure, blood glucose levels and weight) and symptoms once a week.

The nurse typically uses the last one/two weeks of data and simple thresholding to determine critical/non-critical risk. Even this requires up to 15-30 minutes per patient per week.

Our AI-powered risk indicators HyperDetect, CardiacDetect and DiabetesDetect can look into six weeks of data to extend this process to mild and severe stages. As a result, the conditions are diagnosed before the severe symptoms kick in and the cost of care skyrockets.

Our weekly report allows nurses to only focus on patients who have jumped up a category, and as a result, the nurse on average spend 5-10 minutes/patient per week. Since our risk indicators work with existing vitals data in the Electronic Health Record there is no need to change clinical workflows in long term care.

Why does it do beyond state of the art?
1. Only Set of risk indicators that works with data available in the EHRs. This enables us to sell to healthcare providers and health insurance enterprises.
2. Only platform that can tackle weekly multi-morbidity screening in mild stages using AI.
3. FDA approved AI-based risk indicators that completely adheres to FAIR data collection principles (Findability, Accessibility, Interoperability, and Reuse of digital assets).

Healthcare and cost-saving outcomes:
1. In a clinical study with 322 patients at NUGH Galway, we have shown HyperPreventDTx will add value by enabling 41% of geriatric with undetected hypertension [1] reduce the cost of care by €2,350/ patient/ year [2].

Additional market winning features in products:
1. Our Chronic Disease Prevention is also HL7/FHIR v4.0.1 complaint and is embedded in PointClickCare® the largest EHR in the long-term care marketplace.
2. Our platform is also integrated with EHRs like Cerner, eClinicalWorks, Athenahealth and Telehealth platforms like Vidyo in the USA and Wellola in Ireland.
3. Our Chronic Disease Prevention Platform currently uses Microlife A6 as a point of care device.

Why is our solution needed?
• There are approximately 50 million patients in Europe with multimorbidity (two or more chronic conditions)
• The cost of chronic disease management in the EU is €700 billion per annum
• 70-80% of healthcare costs are spent on chronic diseases
• Multimorbidity is frequently first diagnosed in critical stages, due to lack of clinical staff time to analyse patient vitals and symptoms every week
• Most models of care across the EU focus on a treating critical stage and a single disease approach and are not adaptive to the needs of patients with multiple chronic conditions
• Most models of care are also repetitive (multiple appointments), inconvenient, inefficient, confusing (e.g., conflicting advice from different clinicians), burdensome, and
potentially unsafe due to leading to poorly integrated and coordinated care (e.g., medication interactions)
• There is a distinct need for easy to use and integrated digital therapeutics that can diagnose multiple chronic conditions (hypertension, diabetes, and chronic kidney disease) in mild stages and administer nonpharmacological treatment for reversal of conditions. There are approximately 50 million patients in Europe with multimorbidity (two or more chronic conditions)

Describe the main ideas, models or assumptions involved – HyperDetect was trained using large clinical data-sets available in the literature. We first randomly split the data into two groups. The first training group was annotated by three clinicians into seven categories of hypertension and five categories of hypotension. Also, to reduce selection bias we made sure that all the twelve categories have at least 5% of the training dataset.

The second testing data set was used to test the accuracy of the HyperDetect risk indicator. The testing data was again independently annotated by three clinicians. The three physicians had 83% correlation, whereas the clinician’s group has 86% co-relation with HyperDetect risk indicator.

[1] Int J Public Health (2014) 59:759–767 DOI 10.1007/s00038-014-0573-7

[2] Mennini FS, et. al. Cost of poor adherence to anti-hypertensive therapy in five European countries. Eur J Health Econ. 2015 Jan;16(1):65-72. doi: 10.1007/s10198-013-0554-4. Epub 2014 Jan 5. PMID: 24390212.