How do you recognize symptoms that are undetectable by sensors?
Loewenbrück: They are detected by analyzing routine smartphone use for example. The idea here is not to track any websites patients visit but whether there are temporal fluctuations in smartphone usage. All patients undergo longitudinal monitoring where we compare the data of the individual patients. It’s easy to imagine that an impulse control disorder can cause the patient to use his/her phone non-stop. However, the number of phone calls is likely to decrease if the patient is depressed and he/she may stop using messenger services altogether. The project objective is to find methods that collect relevant information but are manageable and realistic for patients at the same time. An alternative would be to send all patients a questionnaire each day, asking questions like "Do you have problems controlling your behavior?" However, patients would likely not want to participate. If we use questionnaires to monitor the disease, patients may be willing to respond only once a month. Incidentally, we will also go this route because there are many validated and tried and tested survey tools available in this area. Patients are asked to regularly complete them via an app. This is ultimately about striking the right balance between high data quality and patient acceptance and the best approach to achieve it.
How do patients know when they should wear the motion sensors? Are there any guidelines?
Loewenbrück: There are reference points of preceding similarly designed projects with which we work in close collaboration. Needless to say, it always depends on what you want to identify. For some parameters, you may have to average data over a longer period of time to obtain clean data. This always depends on the power of the technology, such as battery life and type of sensors for example. The goal is to automatically prompt patients via an app to routinely use specific modules of our telemedicine approach, either by wearing different sensors, completing questionnaires or by performing active tests using a smartphone or tablet.
The sensor-collected parameters are automatically transmitted to physicians. On average, how long does it take from parameter transmission to the analysis/documentation of the results?
Loewenbrück: You have established gait parameters, for example, which can be extracted from the raw data that has been collected by the sensors. Ultimately, the length of time depends on how long you have to collect the data to be able to extract the parameters. For example, if data averaging requires several hours or even days, we will obtain a result after this time. This is always dependent on the examined parameters. The idea is that the parameter data is continuously and as quickly available as possible. Patients exhibiting the highest intra-individual standard deviations should be automatically prioritized and signaled to doctors.