In part one of this series, I shared my experience attending the HIMSS 2025 healthcare event in Las Vegas which was filled with discussions around the growing adoption of AI in healthcare. One of the panels I attended highlighted its applications in administrative tasks. Here are a couple of other examples:
· A triage nurse's role, we are told, is to ask questions and assess risk and then act accordingly. AI predictive algorithms supporting nurses could lead to a better tailored treatment plan. For example, some patients could be allowed home faster than those with multimorbidity, thereby freeing a bed for someone in need.
· AI can also be used when it comes to patient waiting times. By using historical data, scheduling trends, and patient flow analytics, it can forecast wait times. AI can balance capacity by identifying underused facilities, redistributing non-urgent referrals, and assessing clinical data to fast-track urgent cases. Waiting lists can be spread around regions, and patients can select the best provider for surgical and elective care which means they can choose to wait a bit longer for a perceived better delivery.
AI integration needs to be simple, as not all clinicians are computer whiz kids. They are great at patient care; they don't want dashboards to ponder over. They want the answer so they can move on to the next task. And this is a benefit of AI. It simulates understanding to let non-technical users access what is in the box. A panel speaker compares seamless adoption to a small child getting Alexa to play "The Baby Shark Song." If things are easy, behavioural change is easy too.
Healthcare workers are also consumers, and the trends, apps and behaviours we see in the B2C world are being replicated in the workplace. There are wearables for sports training and therapy apps accessible on-demand through a chatbot, quizzes to check info is being absorbed, dashboards to monitor wellbeing, and checklists help structure recovery.
Clinical mobile devices used in hospitals or in the field have sophisticated, enterprise-grade data capture - voice, camera, text - plus they can be equipped with barcode scanners and even radio frequency identification scanning. The growth in data capture and digital health is the perfect environment for more AI innovation.
The ability to record data is important to monitor recovery. Nurses are photographing wounds to check over time. Something I picked up from chatting with customers was that family and friends are sometimes averse to watching clinicians putting data into their phones. It can look like they are playing on them as opposed to attending to an often very sick friend or relative.
I heard about some hospitals resorting to putting posters around to advise they are work tools. Others prefer to go down the route of utilising devices that look clinical. Zebra's healthcare products are blue – designed to highlight dirt so they can be cleaned, but also to ensure the devices look clinical, not like personal devices.
AI is leveraging machine learning models trained on vast datasets of wound images and clinical records. It is already improving wound diagnostics, predicting complications, and optimising dressing materials. But to work, apps, videos and images need space, and processing power. Using the right device designed for healthcare environments with carefully managed data use is an important consideration.
Another possible issue is connectivity, as there is a movement toward greater care in the community. Mobile computers out in the community field – patient homes – can support any number of patient processes, but connectivity is essential. The latest generation of mobile computers are 5G-enabled, but the reality is the 5G infrastructure is still catching up.
Some of my friends in healthcare spoke of areas in deepest rural UK where the network only has 2G. So offline collation must be possible with synchronisation done later. It is important to be able to work offline during these periods and sync the data at the first opportunity.
We also discussed ethical issues around AI. For example, AI models providing recommendations and decisions for things like treatments plans, accountability, human oversight, and the levels of AI model training, quality and trust needed. And no discussion around AI would be complete without a focus on data volume, quality, and the need for healthcare data to reflect the range of demographics within populations.
AI is less likely to put an extra 0 on a dosage or mix up patients, but its output is only as good as the data it synthesises. If it makes a mistake, who is ultimately accountable? And if it does become the perfect mechanism for diagnosing and predicting outcomes, do patients always want to know? Ignorance may be a happier place for some.
For AI to grow into a healthy supportive tool, it is important to develop a culture of transparency, within and across disparate organisations and to learn together. Ultimately, it needs to add value to the patient experience and improve outcomes.
The group also discussed how healthcare is not just about what the doctor says; it is based on a two-way communication with the patient. The patient is the consistent factor as they are there 100% of the time, aware of how they are feeling and able to report changes for better or worse. This is why the 10-minute patient collaboration is a critical part of the treatment process, and one that should not be removed lightly.
In some areas, the patient may even know more than the doctor. Once you or a family member is diagnosed, you immediately head to the internet to learn about the illness, treatment and outcomes (like my family did when my dad was diagnosed with lung cancer.) You become aware of new treatments, drug availability and the wider healthcare ecosystem with the time to focus and connect the dots.
In Conclusion
AI is here to stay. It can help clinicians be more efficient with their time and improve outcomes if it is used carefully, based on good data and synthesised on hardware with the right processing capacity. Transformation is always disruptive but long term, this could deliver better diagnosis, better and more tailored treatment plans plus give time back to clinicians to put back into patient care.
For AI to integrate effectively, it is important to have a governance framework in place with continuous monitoring and feedback to ensure patient safety. Staff must be trained and educated on its benefits and limits. Real world examples are already showing successful outcomes. But the fundamentals remain the same as those for patient care outside of the digital realm. Risk assessments and mitigation strategies must be undertaken for any patient safety-related decisions.
Investing in AI is not a gamble. AI-supported healthcare will be both slicker and smarter, and similar to how the internet changed industries, AI will be the next industrial revolution. The patient needs to be at the centre of everything, and AI must not interfere with this hierarchy. It is there to support –not replace – human engagement.
