Dr Malcolm Pradhan, Chief Medical Officer at Alcidion, and Dr Carolina Garcia Vidal, Head of Respiratory Infection at the Hospital Clinic of Barcelona, discuss how new technologies are making hospitals more efficient in the fight against COVID-19
Author: Beatriz Valero de Urquía
As the shadow of a second lockdown starts to cast its gloom over Europe, the world’s eyes are directed towards the epicentre of the pandemic: hospitals. But, are hospitals’ IT systems prepared to fight the pandemic? Preventable errors in healthcare is the third leading cause of death in the United States. According to a John Hopkins study, medical mistakes lead to 250,000 deaths every year. The number is estimated to be higher — some studies say it’s as high as 440,000 — but the lack of mentions of human error and system failures in death certificates obscures the real numbers. As a response to this situation, a new area of expertise is gaining traction: Health Informatics. Doctors and computer scientists are working hand in hand to come up with technological solutions that improve patient care, reduce medical errors and even predict healthcare outcomes. COVID-19 has only increased the need for these solutions. When the pandemic hit Barcelona last March, the city’s Hospital Clinic was suddenly inundated with patients. Although the hospital has only 800 beds and 10 infectious diseases specialists, by the end of March it found itself caring for 2,000 COVID-19 patients. Carolina García Vidal is the Head of Respiratory Infection and Infection in Severe Immunocompromised Patients at this hospital. She led the efforts to treat COVID patients and manage the infectious disease department during these difficult times.
“What everyone had to do was put all the hospital doctors at the service of COVID patients: surgeons, gynaecologists, ophthalmologists,” she says. “Without them we wouldn’t have made it, of course, but you have to understand that their area of expertise is very far from COVID and that you are therefore placing a huge responsibility on them.” However, since the year preceding the pandemic, Barcelona’s Hospital Clinic had been working on becoming a smart hospital. As part of this project, a team of doctors and engineers were exploring artificial intelligence (AI) solutions to improve medical care. This AI tool was originally being used to support cancer patients with febrile neutropenia, to predict and prevent infections. However, that project was paralysed last March and turned into a Digital Control Centre for COVID-19.
Dr Carolina Garcia Vidal
“At that moment, when we realised how complicated this would be, what we did was put our AI programme at the service of COVID patients," García Vidal says. The AI tool developed by García Vidal’s team is able to analyse over a trillion anonymised data points, identify clinical patterns and predict outcomes, allowing clinicians to provide personalised treatments to COVID-19 patients. The team realised that the complications that COVID-19 patients faced generally fell into one of three categories: complications related to a new infection (co-infection), inflammation or thrombosis. “We realised that each of these patterns was correlated to analytical findings,” García Vidal says. “So what we did was put our AI system to analyse everything that was happening with our patients - 2000 patients - in real-time.” The AI then classified the patients according to how serious their condition was and the likelihood that they would develop a specific type of complication. The tool uses neural networks to analyse the data points and identify the patients who are likely to require mechanical ventilation, or those who might be at an increased risk of dying in the next few hours. The results showed that the AI was able to predict, with 90% accuracy, the trajectory of the disease in a specific patient. The use of the AI tool in the hospital led to a 93.9% improvement in patients’ conditions from the fifth day of the disease and reduced COVID-19-related mortality rate in the hospital by 50%. García Vidal is “extremely proud” of this achievement but also confesses that a significant reason behind its success is that the hospital had already migrated their patients’ entire clinical history to an intelligent system called SILDD (in Spanish, Intelligent System for Reading and Dispensing Data). "The greatest difficulty that hospitals have is being able to read the data in their medical records," García Vidal says. “I think that this is something we are very advanced in, in comparison with other hospitals.”
García Vidal’s team has received a €600,000 grant from the European Institute of Innovation and Technology (EIT) to develop the AI and is now working on expanding the solution to four other European hospitals within the EIT Health network. García Vidal believes that data analytics will be extremely useful in tackling COVID-19 and understanding the nature of the virus, despite the public’s concerns regarding the use of medical data. "I am sure that any doctor who sees these patients improve because of prediction systems, diagnostic aid systems, and treatment aid systems will end up embracing these techniques with total ease,” she says. The main challenge in the digitalisation of medicine is, however, the need to obtain high-quality, unified data that can be transferred between doctors and even hospitals. “In truth, when you are able to create this infrastructure, when you are able to create these intelligent systems for reading and dispensing data, transferring AI from one hospital to another will be really easy,” García Vidal says. “But you have to do that job first. " One company that is helping hospitals “do the job” is Alcidion. Alcidion is an Australian healthtech company that helps healthcare organisations embrace new technologies. Dr Malcolm Pradhan, co-founder and Chief Medical Officer, says that many of the problems that lead to preventable deaths, such as IT failings, have not been addressed in the last 20 years. “When I was going through medical school and then as a junior doctor, I was really struck by how manual everything was, even though we had some basic computer information on the screens,” he says. “Sadly, things haven’t changed that much over the last 20-25 years. People are doing more typing, but a lot of the processes are still the same, and that’s what we are trying to change.”
Dr Malcolm Pradhan
Alcidion was born to provide a new approach to these challenges. “We started up Alcidion to focus on ‘How do we improve healthcare to improve patient safety, to enable clinicians to do their work better?’” Pradhan says. Alcidion is known in the UK for its Patientrack early warning and electronic observations system. Patientrack is currently being used in several National Health Service (NHS) trusts, where it has helped hospitals to reduce cardiac arrests and improve sepsis and acute kidney injury diagnosis. However, Alcidion has recently formally launched another product for the NHS, Miya Precision, conceived as a more comprehensive approach to healthcare and the NHS’ first smart clinical asset. “If you think about current IT systems, they’re really designed as electronic filing cabinets,” Pradhan says. “But Miya Precision has been built from the ground up to be able to examine data as it flows through the system, analyse that data in real-time and apply clinical decision support and algorithms to identify clinical risks, clinicians who forget or haven’t complied with best practice, people who haven’t turned up to appointments who need to be followed-up; all that sort of stuff that is manual and such a cognitive burden on clinicians every day.” Pradhan stresses that the purpose of Miya Precision is not to replace clinicians but to support them by compiling and presenting all the pertinent information so that clinicians can focus on making the best decisions for each patient. “We use real-time natural language processing to assist clinicians as they type,” Pradhan says. “We analyse what they’re typing and recommend best practice guidelines and retrieve data for them so that they don’t have to go manually hunting for it.”
Although Miya Precision uses AI for features such as this language processing system, Pradhan is wary of the limitations of these technologies. “We use AI algorithms more for things like predicting flow and risk,” he says. “But, having been in this area for a long time, I’m very nervous about using machine learning in patient care, because the deep learning models are quite brutal and quite sensitive to noise in the data. And if you know anything about healthcare data, it’s very messy.” Miya Precision also supports healthcare workers by automating manual tasks, such as checking a patient's kidney function or keeping track of the drugs that a patient has received. In addition, Alcidion has launched a remote healthcare capability in Miya Precision and developed Miya Memory, a mobile electronic patient record that enables clinicians to access information anywhere at any time. “Because of COVID, people are accelerating virtual care, which was always going to be the solution in the future,” Pradhan says. “We just didn’t know it would need to be a solution in the first few months of this year, but it was always necessary to scale up healthcare.” In Australia, Alcidion has been using this technology to monitor people at risk during COVID-19 and even worked with the New South Wales local health districts to support the healthcare staff by monitoring patients in their home or quarantine hotels. The pandemic has been a great driver for innovative healthcare solutions, and it has also demonstrated the need for these solutions to be smart and adaptable. “Health organisations had to learn the hard way,” Pradhan says. “Even if you found some good evidence about risk calculation, or a risk algorithm, or even just a good way of managing patients, being able to implement that in any kind of timeframe while under pressure was sure to be almost impossible.” Although hospitals such as García Vidal’s did manage to adapt to the challenges of COVID-19, the Hospital Clinic is unique in its possession of electronic patient records. Nonetheless, it is still unable to share those records and findings with other hospitals. “A big challenge of the Health IT industry is the inability to move data between different systems,” Pradhan says. “And it’s not a technology problem. It’s a strategy by big companies to lock out other competitors.” This is the reason why Miya Precision’s ability to take data from different sources — lab reports, scans, doctor notes, vitals, etc — and standardise it, could be game-changing. “That's a really powerful thing to do because it means that if someone has an innovation in one part of the country and someone else is interested they can much more easily bring it across, download it, and adapt it,” Pradhan says. “That's really a key part of where we want to be: helping clinicians and Trusts improve their care, but also maximise their use of their data, rather than just having it sitting in databases.” Data is one of the most underused resources that the healthcare industry has, and it is fundamental to ensure that there is communication between experts and hospitals. Moreover, in García Vidal’s opinion, it is also an important step in the development of AI solutions. “AI is here to cause a great revolution, to the level of the appearance of antibiotics or other great milestones of medicine, no doubt,” she says. COVID-19 has made the healthcare sector painfully aware of its need to step up its technological game, whether that is through AI, data analytics or automation. There are many challenges on the way, but also many lives to save.