AI is already delivering results in healthcare, but the Philips Future Health Index 2026 suggests health systems are not yet ready to scale the technology safely and consistently.
“Technology alone isn’t enough,” said Carla Goulart Peron, Philips chief medical officer, during a briefing at the Philips Customer Experience Centre in the Netherlands last week. “Without the right infrastructure, AI cannot solve healthcare challenges by itself.”
The annual index, compiled by the global health technology company, aims to better understand how healthcare professionals and patients perceive AI. The 2026 edition surveyed 2,000 healthcare professionals and 20,000 patients across ten countries.
“Those countries don’t represent every healthcare system around the world, but they do provide valuable insights into how AI is being adopted and perceived globally,” Peron told Gadget. “Alongside that research, we’re also conducting more focused studies in specific regions.
“For example, although South Africa wasn’t included in this year’s Future Health Index, we’re carrying out dedicated research across Africa. Just a few weeks ago, we organised a panel discussion in Geneva with representatives from several African countries. The focus wasn’t only on what AI can do, but also on what else is required to make these technologies successful.”
Philips is working with governments, including health and finance ministries in African countries, to understand how AI can form part of long-term national healthcare strategies rather than short-term technology projects.
The Philips Future Health Index 2026 revealed five key insights:
- AI is delivering measurable impact: 71% of clinicians report improved workflow efficiency.
- AI is expanding access to care: 50% say AI has increased their capacity to see more patients, with a global average of eight more patients per week.
- Connected data is becoming a force multiplier: 57% report improved access to consolidated patient data across care teams.
- Clinicians want to move faster than systems: 64% use personal AI tools when workplace options do not meet their needs.
- Scaling AI is constrained by readiness gaps: 70% say training for AI-enabled tools is unavailable, inadequate or inconsistent.
The report found that just under half of clinicians (46%) reported saving time through AI use, allowing them to see an average of about eight additional patients each week.
Peron, an obstetrician and gynaecologist by training, drew on her early experience as a physician in Brazil, where she worked in both public and private healthcare.
She said the difference between the two environments was “remarkable”. In the mornings, she worked in Brazil’s public healthcare system, where every patient had the right to receive care, but resources were often extremely limited. In the afternoons and evenings, she worked in private practice, where access to equipment, time and support differed sharply.
The comparison carries echoes of SA, where many public hospitals carry heavy patient loads while private facilities generally have greater resources and shorter waiting times.
Peron said one case from her early career shaped the way she thinks about health technology. A pregnant woman arrived at a public hospital with an urgent pregnancy complication. She had received limited prenatal care and had not had regular ultrasound examinations.
Although an ultrasound machine was available, Peron said she did not yet know how to use it properly. She performed the examination while holding a telephone to her ear, describing the images to a more experienced colleague who helped her interpret what she was seeing.
That same situation would look different today, she said. The ultrasound images could be shared remotely with a specialist over Wi-Fi or a mobile network, allowing the specialist to guide the examination in real time.
SmartSweep, a Philips solution being validated through a partnership with the Gates Foundation, is an example of technology aimed at widening access. The system is designed to allow people with little or no ultrasound training to perform a guided examination by moving the probe across the abdomen in a set pattern.
The technology does not replace a full diagnostic ultrasound. Rather, the system helps answer whether a pregnancy appears low risk or whether the woman should be referred to a higher-level healthcare facility for further assessment.
The example underlines a broader point in the report: AI can only improve access when health systems have the infrastructure, training and clinical pathways needed to use the technology safely.
Peron gave a second example from stroke care in Brazil. She said mechanical thrombectomy, a procedure used to remove clots from blood vessels in the brain, had strong clinical evidence behind the technology, but Brazil’s public healthcare system was not initially ready to support widespread use.
“There wasn’t an ambulance network capable of rapidly identifying stroke patients,” she said. “There weren’t enough trained physicians. There weren’t enough specialised centres.”
She said the treatment only became available in Brazil’s public system years later, after partnerships between the company, the Ministry of Health, the Brazilian Stroke Society and clinical leaders helped build the required infrastructure.
“Technology is only one part of the solution. It has to be introduced at the right moment, together with the right infrastructure, training and clinical pathways.”
The same principle applies to AI, she said. Healthcare systems need representative datasets, regulation, transparency and human oversight before AI can be scaled safely.
“AI is there to augment clinicians, not replace them. Today’s clinicians manage far more information than they did ten years ago. The number of data points available for every patient has increased dramatically. No individual can process all of that information alone.
“AI helps bring those pieces together. It enables more personalised care. In the past, treatment decisions were largely based on clinical guidelines. Those guidelines remain extremely important. But AI gives us the opportunity to tailor treatment much more closely to the individual patient. Eventually, it may even help personalise treatment based on a patient’s genetic profile.”
Peron said AI is likely to affect healthcare across the patient journey, from prevention and early detection to diagnosis and treatment. Wearable devices already allow people to monitor areas such as activity, sleep and heart rate, while AI can help identify patients at higher risk earlier and draw attention to findings that might otherwise be missed.
In treatment, AI could help clinicians bring together large volumes of information during complex procedures, including CT images, MRI data, ultrasound, physiological monitoring and other sources.
“Bringing all of that information together mentally requires tremendous expertise. AI helps integrate those different data sources into one coherent view.”
She said this could reduce the cognitive burden on clinicians while helping patients gain better access to their own health information.
“I hope that five years from now, conversations between clinicians and patients will be very different,” said Peron. “Patients will have much better access to their own information. They’ll better understand their own health.
“AI is already here. By combining responsible innovation, strong regulation and human oversight, we can use AI to expand access to care and help democratise healthcare around the world.”
* Jason Bannier is a data analyst at World Wide Worx and deputy editor of Gadget.co.za. Follow him on Bluesky at @jas2bann.