Polypharmacy management solution implemented in MedGuide Prototype 1

[by Tudor Cioara, March 2019]

To identify the polypharmacy side effects affecting the daily life activities of patients with dementia, we have addressed the following issues in our prototype implementation (see figure below):

- the detection of significant behavioral deviations from the patient’s baseline out of daily life activities monitored;
- the correlation of these deviations with side effects of drug-drug interactions must be performed.

The first issue is addressed by considering the baseline for a patient with dementia as being represented by the daily activities which the patient is normally carrying out (i.e. daily routine). In the first prototype approach the baseline for a patient is defined by the doctor based on discussions with the patient, its family and caregivers. The information about the patient’s daily activities are provided by IoT sensors. We have considered five types of daily life activities as significant enough for allowing the detection of polypharmacy side effects: sleeping, feeding, toilet hygiene, functional mobility and community mobility. A Random Forest classifier is employed to detect significant deviations from the patients’ baseline considering the monitored activities which may signal side effects of drugs usage.

The second issue is addressed by constructing a Polypharmacy Management Knowledge to capture the potential side-effects of drug-drug interactions and has been used to label historical data. The knowledge base is based on Drug-Drug Interactions Ontology[1] and models the pharmacological effect of drugs, the pharmacodynamics actions of drugs, the mechanisms by which these actions are performed, the processes of absorption, transportation, distribution, metabolism and elimination of drugs, the recommended dose, and the interactions between drugs.

Then, a K-Means algorithm is used to cluster similar days containing significant deviations from the baseline and the results of the clustering algorithm have been used to correlate future monitored days with potential drug-drug interactions. Each cluster will contain similar annotated days, and the label of the cluster is given by the annotation (i.e. drug-drug interaction and its adverse effects) of the cluster’s centroid.

[1] https://bioportal.bioontology.org/ontologies/DINTO

[2] Viorica Rozina Chifu, Cristina Bianca Pop, Tudor Cioara, Ionut Anghel, Dorin Moldovan and Ioan Salomie, Identifying the Polypharmacy Side-Effects in Daily Life Activities of Elders with Dementia, Volume 798 of the Studies in Computational Intelligence series, ISSN 978-3-319-99625-7, IDC 2018. https://link.springer.com/chapter/10.1007/978-3-319-99626-4_33