Clinicians thrive on solving their patients’ health problems
The responsibilities of medical practice can be heavy, but despite the challenges and complexities, or perhaps because of them, clinicians generally find clinical work rewarding. High levels of workplace frustration and burnout currently reported in multiple countries and settings are not intrinsic to the clinical aspects of practice rather they are in large degree, arguably due to operational failures. Operational failures are system failures that impact workflow, patient care, and overall efficiency.

EHRs – cautionary tale of a powerful technology
In the US, Electronic Health Record (EHR) systems were introduced to improve workflow, patient care, and efficiency but have had contrary effects, even during normal functioning as opposed to system “crashes”. The rewards of clinical work are compromised by the frustration of endless data entry and search which steals contact time with patients, instead generating after-hours “pajama time” to catch up on documentation along with a torrent of information out of proportion to the help it provides in managing care.
Is AI the solution to operational failures?
Artificial Intelligence (AI) is a broad term for the newest set of information technologies promising better outcomes. Will AI deliver on this promise, or worsen frustration, get in the way of what patients need and further interfere with how clinicians work? Solve operational failures or create more of them?
Studying operational failures
Mary Dixon-Woods and colleagues in Cambridge have published a valuable piece of research studying operational failures. They used Delphi exercises to gather input from General Practitioners and patients in the English National Health Service (NHS). They identified high-priority operational failures related to EHR issues (e.g., malfunctions, slow performance, or crashes during tasks) but also other causes such as missing or delayed test results (e.g., results not communicated to patients, leading to unnecessary GP consultations), referral problems (e.g., inadequate or outdated referral templates that complicate the referral process), information sharing (e.g., inaccurate or missing information in patient records), and communication issues (e.g., difficulties in contacting other healthcare professionals or patients due to incorrect contact details).
How AI could help solve operational failures
Connecting AI to operational failures might be a route to success in reducing burnout and frustration and restoring joy in work. Let’s look at six types of failure (Figure 1) identified in the NHS study and think about how they could be mitigated with AI.
Test results
AI can improve the management of test results by reducing delays and errors in receiving and communicating test results to patients. AI systems can monitor lab processes, flag delays, and provide alerts about missing results or results that have not been acted upon. Machine learning algorithms can detect discrepancies in test results, reducing the likelihood of errors.
Coding
AI can assist by automatically coding patient information accurately and consistently.
Communication
AI can validate and update patient contact information, improving communication and follow-up care. Predictive analytics can flag potentially outdated contact details, ensuring that communication lines remain open and effective. AI tools can enhance communication with patients, such as preparing email responses and sending automated reminders for appointments and follow-ups.
Administrative burden
AI can reduce the administrative burden related to billing and scheduling. AI-powered practice management systems can automate billing and streamline scheduling.
Referrals and care coordination
AI can reduce inefficiencies and inaccuracies in the referral system that’s often due to outdated or incomplete practitioner information. Up to date, auto-populated referral templates can reduce the administrative burden on GPs and staff, ensuring referrals are accurate and complete, ensuring timely referrals and sharing all relevant patient information. AI systems can track referrals and provide feedback on their status, ensuring patients are followed up appropriately.
EHR transition
AI can aid the inevitable transition from paper to EHR and comprehensive clinical information systems. AI scribes can listen to a clinical encounter and automatically generate a structured clinical note for the record, significantly reducing manual data entry, time and effort. AI can also assist in digitising existing paper records for incorporation to EHRs.

Figure 1. Improving efficiency and patient care in primary care settings by addressing operational failures through AI solutions.
Adapting AI solutions to the South African context: operational failures in South African primary care
Operational failures in primary care, whether in the South African public sector (e.g., in Community Health Centres) or in private practice (solo or small group practices) are likely to overlap with those described above but with distinct differences that should be researched and understood before committing to technology solutions.
For example, private practitioners and public sector clinicians in SA have been protected from the burnout effects of EHR systems; the high costs and technical challenges of EHR implementation have been barriers to adoption. The lack of external mandate or support has also slowed progress. Nevertheless, the absence of electronic records is untenable in the long term and private practitioners are gradually adopting electronic records often as extensions of practice management systems (PMS) that handle billing, scheduling, and other administrative aspects of the practice. Access to capital for infrastructure investments may be facilitated by the recent relaxation of HPCSA regulations governing multidisciplinary practices.

The way forward
Operational failures in general practice and throughout the health system are significant barriers to efficient and effective healthcare delivery. AI offers new tools to address these challenges, but must be tailored to the local context, adapt to infrastructure and resource constraints, and language and cultural diversity.
AI systems will need to be cost-effective, well-supported, scalable solutions that small practices can afford. Practitioners and staff will need adequate training. User-friendly interfaces and comprehensive support systems are essential.
AI offers clinicians the chance to leapfrog to advanced clinical record systems, bypassing intermediate, less efficient technologies. AI-driven EHRs and practice management systems might overcome operational failures, improve operational efficiency, reduce errors, and enhance patient care. Ultimately, they must free up time for practitioners to focus on direct patient care. Both AI and non-AI solutions need to be tested, as technology isn’t always the answer.
Readings
Artificial Intelligence in Primary Care – RACGP
The Royal Australian College of General Practitioners outlines the potential of AI in primary care while also addressing the risks associated with implementation.
Future of Health: The Emerging Landscape of Augmented Intelligence in Health Care
Report by the American Medical Association outlines the evolving role of AI in healthcare, focusing on its applications in primary care, emphasises the importance of integrating AI to reduce administrative burdens and improve patient outcomes, while also addressing concerns regarding bias and privacy.
Exploring Artificial Intelligence and the Future of Primary Care
The transformative potential of AI in primary care, particularly in enhancing diagnostic accuracy and operational efficiency. How AI can streamline processes and improve patient engagement while also addressing the ethical considerations and challenges associated with implementation.