Mental disorders are significant concerns for public health, economic development, and social welfare. A key management issue facing enterprises today is ensuring that they prevent mental disorders, support return to work, and prevent recurrence. For these purposes, we propose the ‘Layered Mental Healthcare’. Mental disorders are originated from brain disorders. The brain disorder causes the physiological changes and affects behaviours. It is important to monitor each of brain, physiology, and behaviour layer, because the impact on each layer depends on individual differences and cases. Also, each care can be supplied for each layer. For behaviour layer, the PC logger and the wristband-type actimeter are used for observing workstyle and lifestyle, respectively. NIRS is useful for brain function and pulse measurement. We can understand a condition, estimate a risk and select a suitable care using these multi-modality data and AI. This system is also helpful for the validations of various cares. We have developed a prototype of a mental health monitoring system with objective biomarkers in each layer. Thirty-nine healthy office workers participated in a trial of the device in an office for four months. Cross validation revealed that the optimum linear models using multi-modality data estimated the scores of “the brief job stress questionnaire” and “K6” with the correlation coefficient of 0.76 and 0.74, respectively. Also, clusters of multi-modality data obtained by k-means had different combination patterns of sub-mood-states. Layered Mental Healthcare system is potentially useful for monitoring mental conditions to prevent mood and anxiety disorders in the workplace.