Reshaping Healthcare Through Artificial Intelligence: Harnessing Data, Wearables, and Real-World Patient Insights for Better Health Outcomes.

Whichever side of the fence you are on, the surging use of, and surrounding interest in, the potential applications of Artificial Intelligence (AI) in accelerating improvements to our health and wellbeing is undeniable. Recent data released by the Office for National Statistics (ONS) highlighted that improved access to healthcare was the most common area in which individuals felt that AI could have a positive impact on their lives, at 31%. While notable variation between respondent groups and compared to the wider population average was seen across other impact areas, such as ‘shopping experiences’ and even the perception that AI would not have a positive impact on their lives, optimism around AI’s use in healthcare remained more consistently at this level, even in those aged 70 and over.

Increasingly, colleagues in healthcare and life sciences are expressing immense interest in exploring ways to utilise the technology available to improve their day-to-day capacity and efficiency, all the while ensuring that patient care remains at an exceptional standard. Indeed, a recent study highlighted that 39.5% of healthcare professionals believed that technologies such as ChatGPT could be useful in medical decision-making processes, as well as patient and family support (44.7%), medical literature appraisal (48.5%), and medical research assistance (65.9%).

Already, AI is supporting across areas such as patient triaging, diagnostics, drug discovery, and bed management – deriving insights from vast sources of data with innovative pattern recognition. So how do we lead the way in best harnessing and accelerating the capabilities of AI in the advancement of healthcare and patient outcomes?

What Does the Healthcare AI Innovation Landscape Look Like

Many of the challenges we face in healthcare today are ones that have seemingly solidified themselves over the past decades. From health inequalities to missed diagnoses and delayed access to vital care, the early steps towards tackling and ultimately eliminating these issues through AI have already been taken – with promising results to date.?

Recently, generative AI, including tools like ChatGPT, have seen a boom in development and usage across sectors and user bases. This form of AI is one based on algorithms that target content creation, be this text, images, audio, or videos, and even synthetic data, simulations, or code itself. Increasingly, this has been applied to the alleviation of clinical capacity pressures and patient support within the healthcare system.

One such innovation is DrugGPT, developed by Oxford University as a means through which to tackle what is currently estimated to be 237 million medication errors each year in England alone. With a dual functionality in reducing medication non-adherence challenges, believed to cost NHS England ￡300 million each year, DrugGPT aims to provide both an additional clinical safety net during the prescribing process and information that supports patient understanding of why and how to take these medicines.

Reinforced by guidance around the underlying research, flowcharts, and references for each AI-generated recommendation, the solution arms healthcare professionals with an instant second opinion based on patients’ conditions, ensuring that potential drug-drug interactions and adverse effects are flagged. Ultimately, such approaches support the blending of AI with a human touch, offering a tool through which to compare clinical recommendations against that of an AI “co-pilot” and equip patients with a better understanding of their medications to drive compliance.

As discussed in some of my recent articles, a recurring theme that has emerged in our work from a patient perspective has been the need for better education and support in understanding their conditions, as well as the potential options available to them. Recognised as a core challenge by the globe’s largest health organisations, the WHO has recently announced a generative AI assistant known as S.A.R.A.H (Smart AI Resource Assistant for Health) to do just that. Designed as a health promotion tool that seeks to provide a digital means of supporting users with risk and causal factors, as well as with lifestyle habits and tips to manage their health, Sarah’s generative AI algorithm ensures that these interactions with users are more accurate, more nuanced, and more empathetic – all in real-time.

At a device level, these AI approaches have also been integrated into remote monitoring wearable technologies to strengthen individuals’ ability to track and influence their own long-term health outcomes. Recently, Oura – a smart ring designed to track key biometrics such as heart rate, body temperature, blood oxygen, as well as sleep and activity – has begun to apply advancements in AI to generate new health insights from the data captured through its device. While not currently FDA-cleared, these include experimental features such as a ‘Symptom Radar’ for early signs of physiological strain, temperature and heart rate-based illness detection, period tracking and prediction, as well as stress detection. Such features in devices like Oura are designed to provide insights that empower patients in their own health, detecting potential issues to drive patient identification of signs, triggers, and potential means of mitigation.

Each new innovation and pathway of real-world implementation strikes a promising chord for health systems and the people they care for. Driving new ways to support the population anywhere from preventative changes in the context of avoidable conditions, to maintaining patient health with better access, therapies, and home-based support, the wider-spread use of AI in ensuring better health outcomes seems closer than ever.

Turning Technology and Potential into Real-World Patient Impact

At Sanius, we have been applying these principles and our experience in advanced AI technologies to support patients, clinicians, researchers, and life sciences at multiple points in the pathway.

This has seen us work with clinical colleagues to establish programmes that provide longitudinal monitoring for patients with chronic conditions, such as asthma, who are at risk of acute and potentially life-threatening health deterioration. Applying medical expertise to the curation of key remotely trackable markers, these are captured through wearable devices and fed into AI models that alert patients and their clinical teams to a potential upcoming need for escalation.

In a similar vein, much of our focus has been on applying Machine Learning (ML) approaches to generate predictive models designed to alert patients with sickle cell disease (SCD) to a potential upcoming pain crisis (vaso-occlusive crisis, VOC). This early predictor AI algorithm has to date demonstrated some successful pilot results, an extract of our work along the development pipeline presented at last year’s American Society of Haematology (ASH) Meeting. With patient feedback already highlighting positive changes regarding early steps to prevent or reduce the severity of their VOCs based on this early warning system, our hope is that this will improve care for many more patients over the coming years.

At a provider level, our AI-driven solutions have focused on supporting clinical teams on the ground with better visibility of how their patients have been faring outside of healthcare contacts to support treatment planning, while also helping to manage patient flow and drive clinically appropriate discharges through advanced AI-driven data integration and analytic visualisations. This support follows through once patients are discharged from hospital to help them stay well at home, including better adherence to treatment plans through enhanced patient engagement, treatment reminders, feedback around correct medication usage, and integrated longitudinal data at their fingertips to track the resulting real-world impacts on their health each day.

Finally, a key ongoing focus across multiple rare disease areas is the targeting of service improvement and a reduction in the current diagnostic odyssey. These are conditions with often non-distinct symptoms that can see patients left without a diagnosis and on inappropriate care pathways for a decade. Our AI research therefore seeks to find patients who may be lost within long journeys to a definitive diagnosis – utilising this technology to identify key common markers along the way and applying this to the wider population in order to detect potential patients much earlier, and with better outcomes.

Each programme and new disease area we embark on with our clinical, patient group, and life science partners is focused on careful curation and defining of the core needs and goals, specific to ensuring the best outcomes for the patients our applied AI is designed to support. With a strong emphasis on ensuring that our AI-driven solutions are usable and have true real-world impacts as they are integrated into clinical team activities, care pathways, clinical trials, research evidence generation, and patients’ day-to-day, we welcome anyone who wishes to learn more about our work in healthcare AI to reach out at [email protected] or speak to our team at the upcoming Haematology Patient and Carers Congress (HPCC24) this May.