In-hospital suicides are often associated with occurring in the evening and during night shifts when there is reduced staff supervision. During these times of high risk, suicides occur in isolated areas of the ward such as bathrooms and single rooms. In this study, a prototype for a Remote Patient Monitoring System (RPM) system was developed for early detection of suicidal behaviour in a hospital based mental health facility.
- Two radio frequency identification (RFID) reader antennas and stable passive tags for data collection were used.
- The reader-antennas were installed in different positions in a research laboratory.
- Along with received signal strength indicator (TRSSI) data, the distance between reader-antennas and tags were measured as Distance_1 and Distance_2.
- Statistical correlation between the features were derived using linear regression.
- Artificial Intelligence machine learning models were used to estimate optimum positions of the reader-antennas for maximum received signal strength indicator (RSSI).
- Decision Tree, random forest and XGBoost algorythm models were implemented to predict the optimum positions.
- Ensemble machine learning approach was employed to extract weighted average of the machine learning models.
