Background and aim
Medication non-adherence impacts patient health and wellbeing, whilst also influencing the safety, efficacy, and cost-effectiveness of medicines. The utilization of mobile health technologies, including text messages, telephone calls and smartphone-based apps, for improving medication adherence (initiation, implementation, persistence) is a growing field, but further research is warranted. This thesis aims to determine the feasibility of using a digital health intervention to support patients in adhering to their medications.

Methods
A systematic review of the literature was conducted to determine the effectiveness of mobile health interventions for improving adherence to oral anticoagulants. PubMed, Embase, Cochrane Central Register of Controlled Trials, and Web of Science were searched from 1/1/2000 to 11/11/2022, identifying 2,319 potentially relevant clinical trials. Of those, 16 studies were included in the final review, and assessed for risk of bias using a draft version of the risk of bias instrument for interventional adherence studies, and quality of reporting for adherence measurement and analysis. A narrative synthesis of the results was presented. Subsequently, a systematic methodological approach was undertaken to assess the feasibility of using artificial intelligence (ChatGPT-3.5) to develop health-promoting messages. Lastly, as part of an ongoing service evaluation by Betsi Cadwaladr University Health Board, the feasibility of a novel mobile health intervention (the Atom5™ app) for supporting poly-medicated patients recruited from community pharmacies in adhering to their medications was evaluated. The app provided medication intake reminders, combined with daily digital messages and gamification in the form of badges, in two languages (English/Welsh). Surveys created in Jisc were distributed to pharmacists across Betsi Cadwaladr University Health Board to gather their opinions of the messages before they were finalised for the app.

Results
Seven of the 16 studies included in the systematic review reported statistically significant improvements in oral anticoagulant adherence, of which four utilized telephone calls or text messages. However, most of the included studies were of poor methodological quality, with 15 studies being at either a high or serious overall risk of bias. Of the 300 initial adherence-related messages generated using ChatGPT-3.5, 108 duplicates were removed, with a further 47 messages removed after screening. This produced a final, refined list of 145 messages. Lastly, within the feasibility study, a total of 10 participants were onboarded onto the Atom5™ app, with 6 weekly adherence, and 3 patient experience questionnaires being completed. All responses to the participant experience questionnaire rated the reminders and digital messages as being useful, whilst also agreeing that they would recommend an app like Atom5™ in the future.

Conclusions
Together, these findings suggest that digital health interventions, including telephone calls, text messages, and smartphone-based apps, may be feasible options for supporting patients in adhering to their medication. Furthermore, our findings allude to the feasibility of using artificial intelligence to generate health-promoting messages.