Authors (including presenting author) :
LEE T(1), KWOK PH(1)
Affiliation :
(1)Occupational Therapy Department, Pok Oi Hospital, NTWC
Keyword 1: :
Artificial Intelligence (AI)
Keyword 2: :
E-Documentation
Introduction :
In occupational therapy (OT) services for in- patients, initial assessment documentation follows a highly structured yet repetitive format. Therapists previously used CMS with static templates, requiring extensive manual deletion, addition, and re-typing of content for each case. This process averaged approximately 10 minutes per new referral, resulting in typos, inconsistent abbreviations, calculation errors in assessment scores and variability in report quality across therapists. The Hospital Authority's phased e-Documentation rollout, commencing late 2024 in Pok Oi Hospital maintained these inefficiencies and reducing time available for patient care in high-volume settings (15–20 new referrals daily per cluster).
Objectives :
To develop "POH OT E-DocuMentor"—a zero-installation, browser-based smart documentation tool created with AI-generated HTML/JavaScript. The tool supports intuitive click-based data entry, real-time automated scoring of standardized assessments (Modified Barthel Index [MBI /100], Abbreviated Mental Test [AMT /10], Hong Kong Montreal Cognitive Assessment [HK-MoCA /30 with percentile], Lawton Instrumental Activities of Daily Living [IADL /27]). Thanks to HATeams webpage, the POH OT E-DocuMentor supports instant generation of a fully standardized, professional narrative progress note ready for direct pasting into the CMS. It is instantly accessible via shared links or QR codes on hospital PCs, CMS stations, Computers on Wheels (COWs), and iPads.
Methodology :
A responsive, tablet-friendly single-page HTML5/JavaScript application was hosted on HATeams (requiring no IT installation or maintenance). It features large clickable elements (checkboxes, buttons and pulldown list), auto-calculation of scores with auto-generated interpretive comments, and a rule-based natural language generation engine for consistent, grammatically correct clinical narratives. The tool is universally accessible by opening a bookmarkable link or scanning a QR code on any hospital device. Developed iteratively in collaboration with frontline OTs and validated for scoring accuracy and report fidelity on real cases.
Result & Outcome :
Mean documentation time per initial assessment decreased from 10.1 ± 2.3 minutes to 48 ± 12 seconds (>90% reduction), enabling true bedside completion. Automated scoring eliminated manual calculation errors entirely. Over the first six months, the tool supported >5,000 initial assessments across Pok Oi Hospitals, saving an estimated 750 clinical hours in 6 months. Standardized report formatting and language virtually eliminated typos and non-standard abbreviations. Access via links/QR codes on hospital PCs, CMS stations, COWs, and iPads minimized workflow interruptions, enhanced handover speed and consistency, and significantly reduced cognitive load. Conclusion: The POH OT E-DocuMentor demonstrates the power of generative AI to dramatically reduce OT documentation time and administrative burden at zero cost and with no IT dependencies. By automating repetitive editing, scoring, and narrative composition—and providing seamless, instant access across all hospital devices via simple links or QR codes—it delivers superior accuracy, standardization, and efficiency. Now established as routine practice for initial assessments in the Pok Oi Hospital, this scalable innovation is well-positioned for broader cluster adoption in OT departments of NTWC, aligning with global efforts to use AI-assisted tools to reclaim clinician time for patient-centered care.
