Enhancing patient safety through AI

By Mohammed Alhajjy, Medical Informatics Specialist, Quality Informatics – Healthcare Information Technology Affairs – KFSH&RC, Riyadh, KSA

Year after year, the world is witnessing discoveries and trends in diseases, treatments, drugs, and technologies related to healthcare. This is evident through the significant advancement in hospital systems, models, and services. Therefore, many governments in the past few years have focused on reforming their systems to adapt to these new trends and to shift from a volume-based system towards a value-based system. However, decision-makers are concerned with the growing demand on healthcare services that are contributed by several factors such as ageing factors, an increase in population, and chronic diseases.

The World Health Organization (WHO) has estimated that 50 per cent of the global burden of disease is chronic illness. These factors are overwhelming healthcare resources within the available capacity and may lead to poor quality of services and result in safety incidents.

This would be resolved if we consider Health Information Technology (HIT) as a remedy that would ensure adherence to quality of standards within safety measures and maintain the minimum requirement.

Technology can help us in shaping our future healthcare system effectively and efficiently by providing the required access to care based on the availability of services promptly. It could retrieve, store, detect, and recommend on behalf of humans as a decision support system. This would ensure patient safety by building up a vigilant system that detects and predicts near misses. For example, administering a Naloxone injection (which is used to treat a narcotic overdose in an emergency) should be reported in a safety reporting system to alert the organisation about such a human or system failure and to establish a system to prevent it in the future. Yet, this incident might be not reported due to several reasons such as safety culture issues due to lack of organisation support in reporting near misses or potential harm occurrences. Alternatively, it could be that the involved provider was too busy to meet other patients’ needs and forgot to report it, or that the reporting mechanism itself might be not accessible or available. In this case, building up rules and queries in the system database and integrating them between the Electronic Medical Record (EMR) and reporting tools would overcome this issue and report it automatically based on the given formula.

This will also support the organisation by providing them more opportunities to learn not to blame reporters upon reporting such potential or actual occurrences to avoid them in the future. Moreover, it should be noted as per researchers that only 10 to 20 per cent of errors are reported. Hence, using technology to automate reporting of triggers could shift to a higher level by inducing quality standards along with safety measures into automated workflows and lead to predict and prescribe through Artificial Intelligence (AI). This shift in healthcare from manual to electronic has reached beyond clinical documentation or merely prescribing orders towards predicting and providing recommendations to what healthcare providers should perform. This is essential to guide the care delivery based on recent validated evidence embedded into the clinical workflows and empowered by technology features in retrieving, storing, and alerting clinicians about their interventions and plans.

Yet, the human factor is crucial and won’t be eliminated but it will be utilised to add more values to their performance. However, as per James Reason, humans tend to make errors intentionally such as mistakes and violations or make non-intentional errors such as slips, and memory lapses. Thus, human errors would be minimised by using technology that will play a significant role in building up a reliable system and eventually will mitigate the risks of both human factors and system failures.

Healthcare is a complex system and its workflows cannot be replicated with every single patient due to their special needs and requirements, but AI could help avoid duplications and overlapping roles among healthcare providers by ensuring well-established communication channels and coordination platforms. This will require healthcare systems designers to consider a LEAN concept while designing workflows and re-engineering processes to ensure the best fit for their organisations and patients. Ultimately, integrating quality and patient safety standards in every single activity performed in designing health IT systems is vital to ensure the benefits of implementing technologies and solutions successfully.

Alhajjy will be giving a special address on “AI and nursing” on October 26, day three of the Nursing Conference, at Patient Safety Middle East.