Predicting Enuresis Treatment Outcomes with the Help of AI

Sammanfattning: 15% of all five year olds and 5-10% of all 7 year olds struggle with bedwetting, a condition medically known as enuresis. Current treatment methods rely primarily on the family to carry out an 8 week treatment. In order to involve healthcare providers in the process and create an opportunity for individualized treatments, Pjama AB has developed an enuresis alarm system. This system directly shares patient data with nurses through a patient portal. Throughout this project, data collected employing Pjama alarm systems is analyzed with the help of AI classifiers in Python. Classifiers when trained on patient data curated predictions concerning the outcome of a patient's enuresis treatment with an accuracy of up to 90%. Further analysis needs to be done on predictions made with trained classifiers in order to reveal at which point in the treatment predictions are reliable. However early observations indicate that this point lies somewhere around 2 to 3 weeks into the treatment. Previously information surrounding treatment outcome could only be concluded after the 8 week treatment period was over. Being able to foresee an outcome of a patient treatment after roughly 2-3 weeks not only involves healthcare providers in the treatment process but simultaneously allows them to individualize patient treatments. In the case of unsuccessful patients, treatments can be halted at 3 weeks instead of 8 saving families weeks of sleepless nights. The hope is that overtime, individualizing treatments early in the treatment period will increase the number of patients which reach a successful outcome and simultaneously make the treatment process easier for families.