Authors (including presenting author) :
Mak HKC(1), Lam SMA(2), Li YC(4), Tam JH(4), Cheung YM(5), Leung AL(5), Lau CKA(6), Cheung WW(6), Lee HWH(6), Ng YKT(6), Wong HSS(6), Chu TKC(3), Tsoi, KFK(3)
Affiliation :
(1)Department of Neurosurgery, Queen Elizabeth Hospital, (2)Department of Medicine and Therapeutics, Faculty of Medicine, The Chinese University of Hong Kong, (3) Jockey Club School of Public Health and Primary Care, Faculty of Medicine, The Chinese University of Hong Kong, (4)Department of Family Medicine and General Out-Patient Clinics, Queen Elizabeth Hospital, (5)Ambulatory Care Clinic, Queen Elizabeth Hospital, (6)Department of Information Technology, Queen Elizabeth Hospital
Introduction :
Despite input of blood pressure (BP) data being available in clinical management system (CMS), limited data was observed. A BP digital monitoring solution was developed for BP capturing, storing and monitoring. The solution includes image recognition of BP readings from BP monitor with the application of edge-computing and cloud-computing techniques. With the solution, it is hoped to improve clinical service for BP measurement and reduce burden of clinical staff on data collection.
Objectives :
We hereby present a clinical service model on BP measurement and a pilot trial on the adaptation of the solution within Hospital Authority.
Methodology :
A pilot trial was conducted in Queen Elizabeth Hospital (QEH) staff clinic. The set-up included an automatic BP kiosk and an office-owned mobile phone. The phone was fixed at a position above the results panel such that the camera could capture the results. Attending patients scanned the barcode on appointment slip received during registration before proceeding to usual BP measurement. After finishing measurement, the phone automatically captured the BP reading by image recognition technique and data was synchronised to workstations of QEH staff clinic directly. Throughout the process, the BP measurement was carried out unattended, with voice prompt for better guidance. Accuracy of image recognition was assessed by validating 10% BP data against photos captured on random selection.
Result & Outcome :
From 1 October to 29 November 2021, a total number of 1114 BP records were collected from 1057 subjects. The mean systolic BP (SBP) and diastolic BP (DBP) were 128 ± 17 and 73 ± 12 respectively. Using office BP measurement of 140/90 as cut-off, 23.5% of subjects were defined as hypertensive. Accuracy of BP image recognition was 100% (111/111). It was estimated that 18.6 man-hours were saved throughout the trial.
