Bipolar disorder (BD) is a mental illness that causes dramatic shifts in a person’s mood, energy, and ability to think clearly. People with BD experience high and low moods — known as mania and depression — which differ from the typical ups and downs most people feel. Impact of bipolar disorder: manic episodes that can be very disruptive and lead to conflicts with family, friends, and society. During a manic phase, people can become aggressive or impulsive. It can lead to risky behavior, relationship breakdowns, financial problems, and even legal issues if someone behaves erratically as part of their condition. People with bipolar disorder have a 15(!) times greater risk of suicide than the general population. It accounts for up to 25% of all suicides. Bipolar disorder has other health consequences and can often occur alongside other conditions such as diabetes, anxiety, cardiovascular disease, and drug or alcohol abuse.
Current conventional approaches include long drug therapy and hospital observation. They have a lot of associated side effects - some medications can increase the patient's appetite or cause changes in metabolism leading to a weight gain.
Our approach is based on a “self-learning system” that “educates itself” and built on each patient regular behavioral patterns. It indicates any further abnormalities in patient life rhythms and state more efficiently. This approach will allow generating specific patient-adjusted daily monitoring. It will also show patient behavior changes. Timely alerts and reactions, including healthcare specialist consultations, may prevent serious treatment disruptions, decrease of overall therapy efficiency, following more drugs prescriptions, treatment timeline, and cost increase.
The fact that system will be built on AI-based analytics using the data markers of each specific patient for “education” allows suggesting usage of the basic set of the hardware sensors as the source of initial data input. These sensors like a heart rate monitor, active movements sensor, pedometer, environmental light sensor, GPS location, daylight time, weekday related activity, and others combined with pattern oriented analysis will give rich data stream to make risk assessment with a precision needed for timely and targeted alerts to the patient, caregivers, and healthcare providers. The benefit of this model lies in decreasing the hardware related cost for end-user and production, extending the range of potential users, and improving the day-by-day usability for the patient.