Diabetes is a growing concern globally, with the number of diabetics being forecast to rise from 422 million in 2016 to 642 million in 2040. As a consequence, the total diabetes-related health expenditure is also set to rise at a rate of 19.2%, reaching € 685 billion by 2040. In Europe, diabetes affects 58 million people and costs € 145 billion per year. There is no cure for diabetes but a proper management of the disease can prevent complications and improve diabetes patients’ quality of life.
Medilync is developing software that employs Machine Learning to maximise the value of the data generated by diabetic patients and convert it into impactful information that helps improving glycemic control. The storage and processing of patients’ data takes place in our cloud backend – Cloudlync. The knowledge thus created is used to empower the patient’s care circle, which is patient-centred and includes caregivers, doctors and other healthcare professionals. Insulync (mobile and web apps) is the interface with users. Besides the basic logging features of commercially available diabetes management systems, we offer a convenient and easy way to upload information from any diabetes or wearable device, possibility for integration with electronic health records, and high quality content adjusted to the needs of the different users. Our objective is to deploy a comprehensive solution for the management of diabetes that increases patients’ engagement and adherence to treatment. We use streaming analytics to deliver real-time insights to the users’ dashboards, supporting patient’s short-term decision making (e.g. the need to compensate for an increased intensity of exercise or carbohydrate consumption). As support for the medium to long term decision making, for both patients and doctors, we developed machine learning models that will be trained and validated with real data from patients. Current approaches deal with limited amounts of information and are not able to account for interpatient variability. Our software will perform an automated case by case analysis of relevant parameters (glucose levels vs. medication, exercise, carbohydrates intake, etc.) to retrieve personalised information on how they interact, affecting glycemic control. It will also be able to predict future blood glucose levels and risk of hypo/hyperglycaemic events in a time frame that allows taking preventive action.
