AdLip: Human Coach-supported Digital/AI Personal Health Assistant to Improve Adherence to Lipid-Lowering Medication

Hyperlipidemia remains as one of the three leading metabolic risk factors underlying AMI onset by 2050. In recent study 3 Asian ethnicities with AMI, the incidence of hyperlipidemia is projected to increase by 205% (341 to 1041 per 100,000 population) from 2025 to 2050. A combination of lifestyle modifications and lipid-lowering therapy is typically recommended for individuals with high LDL-C levels to reduce the risk of CVD. The World Health Organization (WHO) defines adherence as "the extent to which the person's behaviour (including medication-taking) corresponds with agreed recommendations from a healthcare provider"

Poor medication adherence portends poorer health outcomes. In Singapore, around 60% of adults not taking their medications as prescribed (as above) and this creates a considerable economic and clinical burden to individuals and health systems.

The use of digital technology in medication adherence has continued to grow as more healthcare providers and patients recognise its benefits in improving adherence and overall health outcome. Digital interventions have effectively helped patients manage their medication by reminding patients to take their medications on time and providing them with more information about their medications and treatment plan. In the busy world today, the provision of appropriately timed and that perceived to be important would be key to effectively convince intentionally non-adherent patients to take their medicines as prescribed.

This study is a multicentre, open-label, two-arm parallel randomized controlled trial. We intent to randomly assign patients with hyperlipidaemia into one of the two groups: human coach-supported Digital/AI Personal Health Assistant app (intervention group) and standard care (control group) with a 1:1 allocation ratio. The intervention group will receive personalised feedback through the app coupled with human coaching on top of usual clinical care for cholesterol management. The control group will receive usual standard of care for lipid management but will not receive the personalised app nor have access to health coaching.

Participants with hyperlipidaemia (n=376) will be enrolled in polyclinics, and key inclusion criteria are participants who are non-adherent to statins "Extent to Non-adherence" sub-scale of the DOSE Non-Adherence Measure), with a score \> 1 (range from 0-15) with or without on ezetimibe and have LDL-C level above the recommended target levels stratified by risk category. Participants will be followed up at Visit 2 @Month 3, Visit 3 @ Month 6 and Visit 4 @ Month 12 while pill counts will be collected @3m, 6m, and 12m visits. As part of Standard-of-Care, clinical pharmacist will follow-up with patients, titrating lipid-lowering medication (such as statin, ezetimibe etc) as required, and review and take action clinical blood test results.

Only those in intervention group, Human-AI-Health coach will use the information gathered by the AI chatbot to guide the targeted behavioural intervention during phone consultation. The scope of coaching will be strictly related to the medication adherence and general well-being. The coach will not start, stop, or titrate any medication. Coach will escalate concerns to clinical pharmacists when deemed fit. A sub-study of focus group discussion will be conducted with a nested sample of 30-50 intervention group patients. The aims are: (a) to collect insights from intervention patients on their experiences with the app and human health coaching, (b) insights into which intervention components work best for them and under what circumstances, (c) insights into concerns which might impact intervention effectiveness, (d) factors that draw their participation and sustained engagement, (e) factors that deter them from sustainable engagement, (f) factors that may lead other CVD patients to be more inclined to partake in such a intervention and (g) ideas and suggestions to make the intervention more appealing and effective.

• Blood LDL-cholesterol levels [Time frame: 6 and 12 months]
• Cardiovascular Risk score [Time frame: 6 and 12 months]
• Changes in health motivation and attitudes [Time frame: 6 and 12 months]
• Changes in self-care efficacy [Time frame: 6 and 12 months]
• Changes in self-care behaviours [Time frame: 6 and 12 months]
• Changes in quality of life [Time frame: 6 and 12 months]
• Changes in quality of life [Time frame: 6 and 12 months]
• Safety of a new model of care [Time frame: 6 and 12 months]
• Acceptability of a new model of care [Time frame: 6 and 12 months]
• Acceptability of a new model of care [Time frame: 6 and 12 months]