This study investigates the effectiveness and implementation of an AI-assisted rehabilitation tool for adults with frailty in the real-world. The main questions it aims to answer are:

Does AI-assisted rehabilitation among frail adults improve patients' physical outcomes? Is there a particular subgroup of frail subjects that will benefit most from AI-assisted rehabilitation?

Researchers will compare AI-assisted rehabilitation to standard practice (advice and QR link for rehabilitation videos to be done at home) to see if AI-assisted rehabilitation improves clinical outcomes compared to standard practice.

Participants will either:

Undergo an AI-assisted rehab with the AI-sensor tool or a standard practice post-discharge from community hospitals for 12 weeks.

Undergo interval assessments of outcomes. Keep a diary of outpatient rehabilitations (if applicable for the subjects).

This is a pragmatic, randomized controlled study on frail patients discharged from three community hospitals in Singapore.

Inclusion Criteria:

• Adults aged ≥ 21years
• Clinical Frailty Score (CFS) 4 to 6 *
• Provided informed consent (for cognitively impaired persons MMSE<18, next-of-kin or proxy to give consent).
• Have access to smartphone/tablet (own or loaned from investigators)/

Exclusion Criteria:

• Medical conditions precluding safe home exercises or severely affect interaction with AI-tool (angina, vision loss, hearing impairment).
• Cognitive impairment (MMSE<18) and no caregiver to assist with use of tool.
• Participation in structured outpatient physiotherapy (>1 times/week), or day rehab programme.

In accordance to my ethical board and institutional requirements.