Where technology meets politics

The rate at which medical technology and potential treatments are advancing is incredible; the number of medical scientists we currently have is greater than all those who have ever lived. In every country in the world, the medical workforce is intelligent, dedicated, and caring, normally working way beyond their contracted hours for pay below what they could expect in the private sector.

Despite this, real-world outcomes are improving at a much slower pace, and even when there are clear evidence-based guidelines, these are often only completely followed in a minority of patients. The hope is that technology can help us address this imbalance.

I am lucky enough to be a senior leader in the medical profession but also a busy frontline clinician. I hope that this dual lens enables me to be both constructively critical and optimistic, with views that are all my own.

A systems approach to technology

In most large health systems, I feel there are often three significant issues. First is the multiple layers of management that sit within the system. Important decisions often have to be run up and down through seven or eight layers of management; and each layer will generate work, meetings, and requests for more data, all of which are often replicated at other layers, and then within other sectors of the larger network. This is both inefficient and confusing, as different layers will collect different data, making it hard to tease out what is the most reliable and clinically relevant.

Second, those that lead, whether senior system leaders or those within government, often feel they are only visible if they make large-scale changes, although it takes the system many months and even years to adapt, learn, and recover. Leaders change quickly, and new changes are then implemented before the old ones have taken effect and been evaluated. All too often, new ideas are implemented on a large scale, whereas the best way might be to pilot them on a smaller scale, vigorously evaluate them, and then improve the design based on what was learned during the pilot phase.

Third is the push for consistency. Whilst this is sensible in many medical areas, it is less so for technology. Large-scale projects pushing for the same underlying technology are often obsolete within several years, and often fail with enormous costs before they are implemented.

Technology cannot address the second issue, leaders’ desire to do things at scale. This needs a system where healthcare decisions sit outside party politics, with broad stakeholder engagement and strong patient and medical guidance; a system, in other words, purpose-built for making long term decisions. But technology can support this. Currently, most system changes have a poor evidence base and little forensic evaluation. As real-time informatics systems develop, it should be possible to make system change more evidence-based and less political.

Technology, however, can address the other two issues: multiple layers of management and a perceived need for consistency. Evolving and iterative electronic healthcare records, when designed properly, can automatically, and in real time, feed back data on patient flows and outcomes, finance and workforce, thereby removing many of the bureaucratic layers. That might enable better wages for frontline staff, and a more resilient workforce, in addition to the finances to attract higher-caliber nonmedical staff.

Regarding the third issue, the key to looking beyond a “one system for all” approach is the ongoing development of vendor-neutral platforms. If national bodies can agree on the diagnostic and outcome criteria, and the varied and evolving programmes feed into sophisticated data lakes (with inbuilt federated learning, to allow safe industry collaboration) then real-time benchmarking and system learning is enabled.

Technology and partnerships

Health partnerships with either industry or private practice are polarizing, but probably do warrant a pragmatic approach. Where there is clear patient or system benefit, and patient data is completely protected, it should be explored. Medical inflation continues to escalate, and free-to-access health systems will increasingly need larger percentages of the budget, so mutually beneficial offsetting arrangements with transparency and appropriate oversight will almost certainly be part of our future health services.

For industry collaborations, federated learning within an institution allowing joint working and sharing outputs rather than patient data, is a sensible model. Broader collaborations with large organisations also seem sensible as there is a common infrastructure for different areas to feed into, with clear cost savings for all parties. A system that can enable real-time measurements of value-based healthcare (i.e., patient outcomes divided by cost) across a large population can evaluate new models of care or drugs / devices within a real-world environment at a much quicker rate, validate in a different population, and then disseminate the learning nationally and internationally.

Technology throughout the patient pathway

Prevention

The majority of premature deaths are cardiovascular and cancer-related. Lifestyle issues make up around half of this risk, and there are excellent examples of technology improving this situation. Smart watches, home BP monitors, electronic scales, and home sugar and cholesterol kits are already commonplace, alongside a wealth of internet-based advice and support to aid diet, stopping smoking and symptoms to look out for.

The next level of prevention, though, will require studies with big data to look at precision screening; trying to design a personalized approach for when people should consider medical screening such as PSA testing, colonoscopy, CT coronary angiography etc.; helping to define the age at which to start, if at all; and adjusting the frequency of screening as individual variables change. A similar approach is needed for true population health medicine, whereby granular and accurate data can help ensure equitable access and reduce the marked social, geographical and ethnic variations that exist across all aspects of healthcare.

Diagnosis

Currently, it is quite time-inefficient to assess the system. There may be a long wait to access primary physicians, then hospital teams, then appropriate investigations and then follow-up. Intelligent patient access portals are showing some promise with hierarchical questions, potentially on top of home-based simple observations and blood tests, leading to one-stop diagnostic clinics.

Potentially, though, the diagnostic area where technology and particularly AI will become dominant is medical imaging. In multiple areas, AI already outperforms senior clinicians in interpreting X-rays, CT scans and ultrasound images. Studies have shown clear benefits in looking for cancers and infections in CT scans of the brain and lungs, alongside analysing retinal pictures and 12 lead ECGs. AI has also been shown to be superior in some small studies at identifying potential diagnoses based on symptoms and also overall health risks based on electronic health records. As bigger and shared data allow a constant learning cycle, this may well become the default for many diagnostic and imaging modalities, albeit with some senior clinical oversight.

Treatment

There are many areas where technology will aid treatment. Electronic prompts within health records are well proven. Patient resources improve compliance and technology can ensure drugs are actually being taken. Going forward, though, the hope is that we enter a world of precision medicine. Currently, even medically proven advances are ineffective, or even dangerous, in a significant proportion of patients. A health system that can look at patient, socioeconomic, and genomic variables to guide treatment will surely be the future, but will rely on international collaboration and common data collection to allow the patient numbers needed to enable this, and to be resilient as new technologies emerge.

With luck, evolving free text programs and neurolinguistic learning will make the common definitions less important. Genetics is perhaps moving as fast as any area of medicine. Genetic analysis that used to take three years, a room full of computers, and 10 million pounds (US$11.2m) can now be done in 20 minutes on a handheld device. Genetic analysis is mainstream in cancer care, but will rapidly evolve to guide other treatments.

Monitoring

The medical model for monitoring treatment is badly outdated. Typically, patients come back at randomly chosen intervals to see different doctors who are stressed and overbooked. Technology is starting to change this approach, with smartphone monitoring with questionnaires, and home monitoring tools feeding into monitoring systems. At a more advanced level, there are small implantable devices that can monitor heart rhythms and blood pressure 24 hours a day, and automatically update the monitoring hospital. The future will see a technology-enabled approach expanded to most patients who require monitoring, with individually designed questions and monitored observations alongside an easily accessible system for advice, and safety fallbacks if patients can’t or won’t engage. Involving family and friends often improves this type of transformation.

Conclusion

Change is hard for people working within the medical profession, but we will see dramatic changes within the next decade. All too often, the clinicians with the most insight are too busy on their hamster wheel of patients to step off and help design how to use technology, but input from the people on the front line will be vital in ensuring technology is used in a clinically meaningful way. I once sat in a high-level meeting where everyone applauded the complex algorithm that calculated we could take five people off our on-call rota, until the sole clinician in the room pointed out there were currently only four people on the on-call rota.

Clinicians shouldn’t be frightened of technology. It will replace some of the roles of the medical profession, but if embraced will improve outcomes and reduce costs. The key will be organizations and countries working together to enable big data and to share systems and learning whilst completely protecting patient data. That high-level shared vision and joint working will be the key to the medical world realizing the opportunities technology has to offer, which could transform patient outcomes.

Professor Gerald Carr-White is a senior consultant within the NHS in the United Kingdom. He is a consultant cardiologist and clinical director of both a large cardiovascular unit and a wider clinical network. He is a Professor within King’s College London working within the country’s largest AI centre and is vice chairman of the charity Cardiomyopathy UK.