Following earlier articles on consumer digital health services and maturity models, this piece looks inside hospitals: where can artificial intelligence (AI) deliver measurable value in a system whose main constraints are limited capacity rather than lack of technology. Workforce shortages, rising patient volumes, overloaded services and funding pressure create daily operational challenges; digitalisation and AI can help make existing capacity more effective.
Where the earliest effects are likely to appear
Public discussion often highlights diagnostic breakthroughs, but in the short term AI’s clearest effects are more likely to come from automating administrative and operational tasks and improving information flows. AI’s value lies in how well it integrates into hospital workflows and relieves system bottlenecks, not merely in the technology itself.
A recent WHO overview reports that 74% of EU countries already use AI for diagnostic purposes (for example analysing imaging, disease detection or clinical decision support), and 63% use chatbots to support patient engagement. According to the WHO, in Hungary AI may be especially useful for reducing pressure on health workers and for addressing health inequalities.
Three levels where AI can create value
Viewed as a complex system, a hospital can gain from AI on three levels:
- supporting clinical care and decision preparation;
- improving patient information and engagement;
- making background operational and administrative processes more efficient.
Each area serves different objectives but ultimately contributes to solving patients’ problems and producing systemic impact.
Unburdening clinicians: AI as a digital scribe
One high‑impact application is using AI as a digital scribe to reduce daily operational burdens that do not directly relate to care. AI scribes can summarise clinical documentation, prepare handover notes and discharge summaries, support context‑aware information retrieval, and handle e‑mail communication with patients.
A published study found that an AI scribe produced a 69.1% time saving for total documentation time compared with a conventional, non‑AI workflow. In primary care, after an eight‑week period, note‑writing with the AI tool averaged 4.16 minutes per note—18.59% less than the 5.11 minutes without the tool—and note length increased on average by 5.4%. Another study reported that at least half of participants saved at least 20 minutes over a half‑day period of use.
A Stanford pilot on clinical AI use reported a significant reduction in cognitive load for health workers after adopting AI tools. A JAMIA article also highlighted a significant decrease in the proportion of staff experiencing burnout. These effects can improve both efficiency and patient experience by freeing clinicians to spend more time on direct patient contact.
Decision support in clinical work
AI can act as a digital assistant for decision preparation: faster review of patient histories, clinician preparation before visits, monitoring of drug interactions, optimisation of triage systems, and automated annotation in imaging are examples where AI supports evidence‑based practice. Some sources suggest AI diagnostic tools can match or exceed human accuracy in specific tasks; nonetheless, final clinical responsibility must remain with physicians, with AI serving as validated decision support.
Patient information, transparency and engagement
Patient experience is shaped not only by clinical quality but also by timeliness, clarity and opportunities for involvement. Information asymmetry and fragmented patient journeys are frequent sources of frustration. Generative, professionally overseen AI chatbots could help by explaining discharge summaries and test results in plain language, reducing patient anxiety.
Chatbots and virtual phone assistants can also help with appointment booking that accounts for individual preferences and institutional constraints, deliver timely, comprehensible pre‑ and post‑procedure instructions following professional protocols, and support follow‑up communications. These tools can contribute to personalised preventive advice and to patients taking a more active part in their care.
Operations and administration
The third key area is the hospital’s operational backbone: capacity planning, data coordination and administrative workflows. Many current processes contain inefficiencies that represent development potential; AI can expose and exploit these reserves, amplifying achievable efficiency gains.
Transforming administrative workflows—automated reporting, funding data submissions, billing and contract management—can yield substantial efficiency gains and cost reductions. Automating administrative processes could enable savings of around 30% in some cases. AI contract‑management systems can reduce legal and administrative workload and process historical, even scanned documents via free‑text search, improving access to archived contracts.
Data‑driven capacity planning, patient flow forecasting and automated billing and reporting are relatively quick‑to‑deploy solutions that can deliver tangible results in a short time frame, primarily by reducing administrative and operational workload and thereby indirectly easing clinicians’ burdens.
Implementation considerations and risks
Introducing AI into hospitals is complex and raises technological, organisational and operational questions. Starting with smaller pilots helps institutions assess required process changes and the daily impact of new tools; broader rollout decisions should be based on those experiences. Key considerations include:
- Problem first: AI should be deployed only for genuine operational needs.
- Process redesign is unavoidable: AI changes workflows and those must be analysed and adapted.
- Risk management and human‑in‑the‑loop: final clinical responsibility must remain with health professionals; AI should be a validated support. Special attention is needed to detect and manage hallucinations in generative models, and ethical oversight and clinical supervision are essential.
- Data quality: AI delivers reliable outputs only from good quality data and functioning hospital information systems.
- Professional attitudes and training: workforce education must keep pace with technology so staff understand use, limits and risks; users should apply critical judgement given responsibility and hallucination risks.
Conclusion
Digital maturity in hospitals is ultimately reflected in how well processes are redesigned physically and digitally in an integrated, user‑friendly way. Technology alone does not deliver a digital leap: well‑organised processes, good data and trained professionals are required. AI can add real value when it supports background operations, freeing time, attention and capacity for what matters most: patient care. Less documentation time means more resources available to the patient—the central actor in healthcare.
This article is part of our series on health‑care digitalisation. Earlier parts examine the digital health ecosystem and stakeholders, a maturity model for the domestic ecosystem, and a case study on digital services in medication management.
