It's difficult to identify falls in senior care facilities. Because the resident, who may be suffering from mental illness, is not instrumented by vision-based fall detection systems, it is a major advantage. Fall detection technology are being used in nursing homes as part of this research. A Convolutional Neural Network (CNN) trained to optimise a sensitivity-based metric is the basis of the suggested solution, which is based on Deep Learning techniques. It discusses the medical criteria and how they affect the CNN tuning in this study. The findings demonstrate the relevance of the timing component of a fall. Consequently, the medical team's aims are best served by a specific measure tailored to this use case and an implementation of a decision-making process.