Identification of Patterns in Cystic Fibrosis Physiotherapy with Unsupervised Learning
Healthcare Systems, Population Health, and the Role of Health-Tech Workshop @ ICML Workshop, 2020
Olga Liakhovich, Tempest van Schaik, Bianca Furtuna, Mihaela Curmei, Emma Raywood, Helen Douglas, Kunal Kapoor, Nicole Filipow, Eleanor Main. https://drive.google.com/file/d/1d7R_oRW4IJoczJ8_fSRdOjn9u2cP73sv/view
Cystic Fibrosis is the most common life-limiting inherited disorder in the UK, affecting approximately 1 in 2500 babies born. It is a systemic genetic disorder that mainly affects the respiratory system. Excessive thick sticky mucus can cause cycles of infection, inflammation and lung damage leading to a deterioration in health. Despite improvements in care, Cystic Fibrosis remains progressive and incurable. Doctors routinely prescribe airway clearance therapy (ACT) to patients along with suggesting an increase in physical activity. ACT physiotherapy entails breathing into special devices that promote the elimination of mucus. To date, there have been no controlled studies that validate its effectiveness.
Project Fizzyo is the first such trial which uses sensors to remotely capture and transmit time-series ACT and physical activity data of 160 children over 18 months. In partnership with UCL Great Ormond Street Institute of Child Health and Microsoft we develop a framework for analyzing and discovering patters in the clinical research data.
