Machine learning for objective symptom detection

The use of machine learning to transform raw vital sign sensor data into objective health-related symptoms represents an important advancement for remote patient monitoring. This is especially valuable in women’s health which has traditionally been underserved by healthcare systems. Women’s health needs exist throughout the lifespan and many are not disease-related. Menopause is a life transition experienced by every woman and in the 5-10 years before menopause, known as perimenopause, women experience a broad range of symptoms that negatively impact their quality of life, health, and work productivity. Critically, untreated, these symptoms also place women at a increased risk of chronic disease post-menopause. There is no objective way to diagnose perimenopause and clinicians have to validate self-reported symptoms from women in the clinic. This leads to an underdiagnosis of perimenopause, leading women to require unnecessary visits to clinicians and for hospital procedures, while suffering unnecessarily with symptoms. The objective detection of perimenopausal symptoms is an ideal application of machine learning, based on vital sign sensor data. This approach can utilise the features from different sensors to develop specific algorithms for individual symptoms, rather than just relying on physiological responses, for example heart rate; thereby differentiating between perimenopausal symptoms and other daily events such as exercise. Predictive accuracy is crucial in managing perimenopausal symptoms and machine learning algorithms are capable of continuous learning, meaning they can adapt and improve over time as they process more data. The objective detection of perimenopausal symptoms will lead to earlier diagnosis, better and personalised management, a reduction in the number of hospital-based procedures and, most importantly, empower women with information about their symptoms that will allow them to live healthier and more productive lives, for longer.