May 31, 2021

The rise of predictive models and algorithms in healthcare 

Lucy Mackillop
5 min
Lucy Mackillop, CMO at Sensyne Health, tells us how predictive modelling can benefit healthcare

The last year has seen incredible transformation in the healthcare industry, with technology playing a central role. For example, we have seen amazing research progress at phenomenal speed in efforts to tackle the novel coronavirus SARS-CoV-2; with rapid development of new treatments, rigorously tested in time frames only dreamed of before. 

Technology has also been rapidly adopted to help maintain services while minimising the risk of infection.  As societies reopen, getting hospitals, clinics and research centres back to pre-Covid focuses is the next priority.  

With so many new technologies and processes embedded in such a short space of time, there is also a new opportunity for the healthcare sector to fully embrace innovation with the more widespread adoption of Machine Learning (ML). 

ML techniques have the ability to influence clinical care by utilising the vast qualities of data generated by technology adoption in healthcare, presenting insights from ML algorithms to the clinician informing clinical decision making.

Embracing predictive models in healthcare 

The growing confidence in the power and breadth of these technologies resulted in the UK government pledging an additional £250m to boost the role of Artificial Intelligence (AI) within the health service post-Covid-19. 

With technology as an enabler, medical professionals have the potential to work with greater efficiency, allowing greater time with their patients and more personalised care. One such example of where technology is already having a positive impact, is in the development of sophisticated algorithms that have helped improve patient care during the pandemic. 

ML analysis of large sets of real-world data, such as electronic patient records, can create predictive models and algorithms. One benefit of this is being able to more accurately predict certain health conditions developing in particular patient groups, potentially enabling earlier intervention or identifying treatments that are likely to be most effective for a particular patient. 

By comparing an individual’s healthcare records against a database of millions of other anonymised patient records, clinicians have more information to help inform their and their patients decision making and personalise treatment plans. 

Supporting decision-making

For several reasons, not least that human interaction, not technology, is at the core of the clinician-patient relationship, the sector previously has sometimes been slow to embrace new technologies. However, the use of technology has become widely accepted by clinicians as capabilities develop and evidence of their clinical impact becomes apparent. 

A good example of how this works in practice is SYNE-OPS-1, an operational AI algorithm, launched earlier this year, which provides real-time operational decision-making support to NHS managers who are coping with the pandemic. 

The algorithms offer an hourly risk forecast for the number of patients expected to be transferred to ICU or be put on mechanical ventilators, allowing hospital managers to direct their resources more effectively as patient numbers fluctuate.  

The pandemic has highlighted to clinicians that using modern technology can augment the decision making process and relive pressure on healthcare services – something that is going to be critical if we are to recover from the 4.7 million appointments missed as a direct result of the pandemic. 

Applications for maternal health 

Maternal health is a prime example of where technology has been embraced by clinicians and patients, and is already being used as a force for good in ensuring the safety of expectant mothers and their babies. 

Regular check ups and monitoring are a key part of the care of pregnant women, and with Government advice to limit face-to-face appointments in this vulnerable group during the Covid-19 pandemic, new processes had to be rapidly put in place. 

This ensured women were remaining healthy during pregnancy and surveillance for, for example, for hypertensive disorders that affect 1 in 10 pregnancies worldwide, was able to occur safely.

As with much of our lives now, video calls have become the norm and recent UK research has shown the healthcare industry is no exception. Over half (57%) of those surveyed believe the ability to see a healthcare professional remotely during the pandemic has been important and helpful. And for maternal health, these continued virtual connections have been critical in the ongoing support for women during their pregnancy.  

Predictive modelling and algorithms, coupled with remote patient monitoring, have made it easier and safer for clinicians to identify when specific treatments are needed. They can better predict which women are likely to need medication to control diabetes during their pregnancy, or which women should adopt certain lifestyle measures and take earlier intervention to help them. 

Remote patient monitoring increases efficiency in healthcare, by allowing clinicians to filter patients and change the operational mindset from ‘who are we seeing today’ to ‘who needs attention today’ and prioritising those who are most unwell. 

The future of predictive healthcare 

We are also likely to see the predictive healthcare model applied to other areas of the healthcare sector. Chronic diseases like diabetes, chronic obstructive pulmonary disease and heart failure will, in the very near future, routinely benefit from predictive modelling and remote patient monitoring, not only making it quicker to identify illnesses, but also improving patient care. 

While long term management by remote monitoring is more challenging, specific short-term intensive support can be well addressed for patients suffering with chronic diseases, particularly when new treatments need starting. For example, for patients with type 2 diabetes who are transitioning to using insulin therapy or heart failure patients post discharge from hospital, short term monitoring can provide intensive support and care.  

The last twelve months have seen technology become an invaluable asset for the healthcare sector, more than it has ever been before. As the level of understanding, acceptance and capability of technology increase, the more opportunities there are to make a difference. Although previously the healthcare industry has been slower to embrace new technology, AI and ML are now helping to drive a digital healthcare revolution.

Share article

Jun 23, 2021

Introducing Dosis - the AI powered dosing platform

3 min
Dosis is an AI-powered personalised medication dosing platform that's on a mission to transform chronic disease management

Cloud-based platform Dosis uses AI to help patients and clinicians tailor their medication plans. Shivrat Chhabra, CEO and co-founder, tells us how it works. 

When and why was Dosis founded?
Divya, my co-founder and I founded Dosis in 2017 with the purpose of creating a personalised dosing platform. We see personalisation in so many aspects of our lives, but not in the amount of medication we receive. We came across some research at the University of Louisville that personalised the dosing of a class of drugs called ESAs that are used to treat chronic anaemia. We thought, if commercialised, this could greatly benefit the healthcare industry by introducing precision medicine to drug dosing. 

The research also showed that by taking this personalised approach, less drugs were needed to achieve the same or better outcomes. That meant that patients were exposed to less medication, so there was a lower likelihood of side effects. It also meant that the cost of care was reduced. 

What is the Strategic Anemia Advisor? 
Dosis’s flagship product, Strategic Anemia Advisor (SAA), personalises the dosing of Erythropoiesis Stimulating Agents (ESAs). ESAs are a class of drugs used to treat chronic anaemia, a common complication of chronic kidney disease. 

SAA takes into account a patient’s previous ESA doses and lab levels, determines the patient’s unique response to the drug and outputs an ESA dose recommendation to keep the patient within a specified therapeutic target range. Healthcare providers use SAA as a clinical decision support tool. 

What else is Dosis working on? 
In the near term, we are working on releasing a personalised dosing module for IV iron, another drug that’s used in tandem with ESAs to treat chronic anaemia. We’re also working on personalising the dosing for the three drugs used to treat Mineral Bone Disorder. We’re very excited to expand our platform to these new drugs. 

What are Dosis' strategic goals for the next 2-3 years? 
We strongly believe that personalised dosing will be the standard of care within the next decade, and we’re honored to be a part of making that future a reality. In the next few years, we see Dosis entering partnerships with other companies that operate within value-based care environments, where tools like ours that help reduce cost while maintaining or improving outcomes are extremely useful.

What do you think AI's greatest benefits to healthcare are?
If designed well, AI in healthcare allows for a practical and usable way to deploy solutions that would not be feasible otherwise. For example, it’s possible for someone to manually solve the mathematical equations necessary to personalise drug dosing, but it is just not practical. AI in healthcare offers an exciting path forward for implementing solutions that for so long have appeared impractical or impossible.

Share article