Artificial intelligence in anaesthesia: shaping the future of workforce and wellbeing

Abstract

Burnout is a syndrome characterised by emotional exhaustion leading to frustration, fatigue and a lack of professional efficacy [1]. Healthcare professionals are particularly susceptible [2], resulting from disruption of the delicate balance between workload and factors that contribute to career fulfilment. Factors that influence trainee burnout are well established [3]. Anaesthesia residents can rotate hospitals every 6 months and poorly designed, rigid rotas that lack transparency have been highlighted as a key contributor to burnout [4].

Artificial intelligence (AI) has the power to revolutionise efficiency in many areas across healthcare, but its role in well-being has yet to be considered. This study examines the application of AI in designing a rota for 27 residents in an anaesthesia department. It was hypothesised that AI can assist with the delivery of complete 6-month rotas, equal share of on-call commitments and facilitate flexibility with leave requests. A rota template was constructed using Microsoft Excel (Microsoft Corporation, Redmond, WA, USA) with commands built from ChatGPT (OpenAI, Inc., San Francisco, CA, USA). The programme was tailored for the requirements of an anaesthetic department. The process then began by assigning leave ensuring minimum daily staffing requirements were met and all residents received their leave entitlements. The on-call rota was then assembled across three tiers: operating theatres; obstetrics; and critical care. A weekly rota was designed to auto-populate, accounting for the on-call, post call and leave schedules. The study investigates the performance of the AI-derived rota for two 6-month cycles (July 2024 to January 2025 and January–July 2025).

Residents received a complete 6-month rota before commencing the post. There was an equal spread of on-call commitments accounting for planned changes between call tiers due to training progression. During the first 6 months, the mean number of on-calls for operating theatres, obstetrics and critical care was 23.8 (95%CI 20.3–27.3); 42.5 (95%CI 41.3–43.7); and 37.7 (95%CI 32.2–43.2), respectively. The on-call frequency during the second cycle was 31.5 (95%CI 30.4–32.6); 33.1 (95%IC 32.5–33.7); and 27.4 (95%CI 23.2–31.6), respectively.

Artificial intelligence-assisted decision-making resulted in a turnaround time of 1 day for the final approval of all leave requests. A call frequency tracker was published to ensure rostering transparency. Table 1 highlights the results of the AI-derived rota for the two 6-month rota cycles. The performance of the programme improved following minor adjustments after the first 6-month cycle. Sixty-four (89%) of first preference annual leave requests were approved which increased to 68 (100%) during the second cycle. All residents sitting exams (n = 20) received a minimum 10 days of leave and all attended college training days (n = 25).

The British Medical Association sets standards that a duty rota must be released 6 weeks before commencing a post [5]. Such a directive does not exist in Ireland. On-call rotas are typically issued with minimal notice, sometimes just a week before starting a post. It is common practice that rotas cover short periods, for example, 6 weeks or 3 months. Residents moving between different levels of call tier, rota gaps and fair allocation of subspecialty modular time are often cited as barriers to rota flexibility.

Education leave for college examinations, courses and conferences is allocated typically after annual leave, which can contribute to stress. Given the complexity of rota design, residents are encouraged to request leave in 1-week blocks and up to 6 months in advance. The AI-designed application tracked the number of residents off on any given day over the 6-month period and predicted additional absences such as EU Working Time Directive rest days. These predictions enabled leave requests at short notice and outside of the conventional week blocks. Artificial intelligence facilitates flexibility and may reduce the administrative burden aiding departmental self-rostering with the attendant benefits.

An effective rota should consider all anaesthetists' needs. Residents returning from extended leave, like maternity leave, had customised reintroduction to on-call duties. Forethought that ensures patient safety and reduces stress is essential in rota design [6].

This study has no benchmark for comparison. Although on-call frequency is audited nationally, rota issuance and leave approvals are not scrutinised to the same degree. Departments should audit rotas to assess their impact on resident well-being. Moving away from fixed weeks of leave, limited educational leave and on-call rotas released with short notice may help reduce burnout. Using innovative workforce planning guided by AI could be an effective approach to improve rota design and potentially enhance well-being.