AI has been at the forefront of the medical profession’s efforts to fight Covid-19 and treat patients during the coronavirus pandemic. Enabling healthcare providers to make fast, accurate and data-driven decisions, the technology has been producing some extraordinary outcomes.
Outside the Covid crisis, machine intelligence is lending itself to hundreds of medical applications, from scanning vast numbers of people to assess their risk of dementia to accelerating the drug discovery process. Here is just a small selection of cases where the technology is revolutionising healthcare provision.
1. Easing the admin burden
Healthcare professionals are using AI-powered speech-recognition systems to update electronic patient records more quickly and accurately.
“This has had a big impact on the efficiency of getting our letters done for any clinic,” reports Dr Paul Altmann, chief clinical information officer at Oxford University Hospitals NHS Foundation Trust. “Clinicians can now send letters within 24 hours, or even instantly if they aren’t waiting on blood test results. Before the speech-recognition era, letters took two weeks or even longer to be completed.”
Dr Simon Wallace is chief clinical information officer at Nuance, a provider of speech-recognition systems. He says that clinicians spend an average of 11 hours each week creating documents. If you factor in time wasted on lost and repeated documentation, they could be spending up to half of their working week on admin – a lot of time to be away from their patients.
“We speak three times faster than we type. By using AI-powered speech recognition, clinicians save time on documentation processes, which enables them to spend more of it on delivering care,” he says. “Even when clinicians talk extremely quickly at length, speech-recognition technology can recognise and record all their words, transforming them into detailed medical notes. Thanks to the capabilities of the software, all medical terms are automatically recognised in the correct context. The only thing that users need to do is use their voice.”
Wallace believes that this facility could significantly reduce the admin burden associated with remote consultations and could even help to reduce the risk of burn-out among NHS staff, many of whom have been stretched to the limit by the Covid crisis.
2. Improving communications with patients
Conveying information to the public in a timely and effective way has been crucial during the Covid-19 vaccination programme. It’s therefore vital that they don’t ignore important messages. This was the challenge faced by pharmacy chain Walgreens as a vaccine provider in the US.
Brian Tyrrell, its senior director responsible for customer marketing platforms, stresses the importance of the wording of the company’s messaging. “You’d expect people to be more responsive to an email telling them critical health information, of course, but we had a duty to ensure that our messages were heard,” he says.
Walgreens, which can reach about 50 million people via email, has been using Phrasee’s AI technology. The system learns with each communication the type of language that connects best with the target audience. “This means that we can double down on our efforts to reach the maximum number of people with relevant updates as time moves on,” Tyrrell says.
Walgreens estimates that 30% more of its email recipients than normal have opened a message from the company since it started applying the Phrasee system to the vaccination campaign. Potentially, that could mean up to 30% more people than normal reading information about appointments – and up to 30% more getting vaccinated.
3. Monitoring for adverse events
Agencies such as the US Food and Drug Administration and the UK Medicines and Healthcare Products Regulatory Agency have systems in place to monitor reactions to any drug, device or therapy. Biopharma companies also have their own systems for monitoring adverse events for their products. These are under more pressure than normal owing to the Covid crisis.
Dr Vladimir Makarov is consultant and programme manager at the AI Centre of Excellence at Pistoia Alliance, a not-for-profit body advocating greater collaboration in healthcare and life sciences. He says that AI supports the monitoring process, “which will be straining under the extra workload imposed by the worldwide vaccine roll-out. AI also helps to interrogate data and evidence from various healthcare environments, from understanding the effectiveness of a drug in a real-world setting to analysing genome sequencing.”
AI is also being used to identify existing treatments that can be repurposed for patients. It gives researchers an efficient way of reviewing large volumes of data and uncovering new insights. For instance, the Pistoia Alliance recently collaborated with a group of partners on an AI ‘datathon’ for drug repurposing. In one month, this helped scientists to discover four repurposing candidates with potential to treat chronic pancreatitis – a disease affecting about 1 million people worldwide – which currently doesn’t have an approved treatment.
4. Automating Covid-19 reports
The pandemic has presented many new challenges, one of which has been the introduction of additional daily reporting responsibilities.
As part of the NHSX AI Lab project, tech consultancy Foundry4 took on the task of automating the daily Covid situation reports being sent to NHS England by Kettering General Hospital NHS Foundation Trust.
“A lot of our time was taken up with conveying an accurate picture of how the hospital was coping,” says Ian Roddis, the trust’s acting digital director. “Although it was a vital part of the national response, we were adamant that the process would not prevent us from dedicating as much of our time as possible to patient care.”
He and his colleagues were aided in their efforts by the introduction of the Mary Bot, named after the trust’s head of IT. This is software that autonomously draws data from patient and HR admin systems to help manage the reporting process.
“Removing the heavy administrative burden from our clinical staff has not only enabled them to spend more time attending to patients; it has also eradicated errors in our reporting,” Roddis says. “It’s ensured that we’re providing an accurate picture that supports that national pandemic response.”
On top of this, the automation has saved an estimated 4,400 working hours a year, representing an annual cost saving of more than £150,000.
5. Reducing stroke recovery times
Stroke survivors are often left with long-term health problems such as impaired mobility, which can require periods of rehabilitation lasting months or even years. Fewer than a third of patients fully regain their mobility and strength.
ReLive, a start-up based at Nazarbayev University in Nur-Sultan, Kazakhstan, is working to improve stroke patients’ recovery times with the aid of rehabilitation robots. These devices are wearable exoskeletons that can be strapped around the affected limb. They are controlled using software that reads and interprets the patient’s brain signals via an electroencephalogram (EEG). Information about their progress can be recorded during the rehabilitation exercises on an online platform for clinicians to analyse.
The system can interpret the signals obtained by biosensors, such as an EEG, and pass these on to the robot controller. It can also learn an individual’s gait to fine-tune the robot’s movements and optimise these so that the stress on the soft tissues surrounding the affected limb is minimised.
6. Detecting the early warning signs of dementia
AI-enabled systems have come to the forefront in the diagnosis of Alzheimer’s disease and other forms of dementia. For instance, they can note slight changes in speech, such as the elongation of pauses between words, a growing preference for pronouns rather than proper nouns and the use of overly simplistic descriptions.
ViewMind is one company that’s using AI to detect Alzheimer’s and similar conditions at the pre-clinical stage. Using a VR-type headset in a test that takes about 10 minutes, it’s aiming to assess 1 billion people a year in settings such as GP surgeries and opticians as part of a standard health check. The system captures more than 10,000 data points through eye movements. The company says that this can detect cognitive problems 20 years before the symptoms present themselves.
ViewMind recently conducted a study that has found a significant portion of Covid-19 patients have a form of so-called long Covid that could trigger potentially serious cerebral side effects. The researchers used the headset device on patients to stimulate parts of their brains and measure their eye responses. The results were fed through an AI system to analyse the cognitive impact of the virus.
7. Improving fertility treatment
Virtual fertility clinic Apricity uses a proprietary algorithm to predict its patients’ chances of successful conception, with and without treatment.
“Our patients have found great value in using our AI fertility predictor tool, which analyses lifestyle factors to evaluate the likelihood of pregnancy,” says Dr Cristina Hickman, Apricity’s chief scientific officer. “It can suggest different types of treatment to support them in their quest for fertility.”
She continues: “Our team is developing machine-learning models with applications in a number of areas, including embryo selection, follicle-scan interpretation and hormonal treatment. When we’re culturing embryos, for instance, we can capture many images and millions of data points. We can use these to create an algorithm-based standard classification, where we assign a score to each embryo and predict its chances of becoming a healthy baby. Compared with traditional methods, this would save the time, money and stress usually spent on unviable embryos, resulting in more babies per patient and better standards of clinical practice.”
AI can also assist the clinicians in standardising decisions such as choosing the best sperm and determining the optimal number of embryos for transfer to maximise success rates while minimising multiple births.
“Ultimately, AI can help to involve patients in the key decisions,” Hickman says. “This brings simplicity, autonomy and less stress to a journey that, without AI, is anything but straightforward, controllable and stress-free.”
8. Accelerating pharmaceutical R&D
Machine learning is proving invaluable to the pharma industry in the drug discovery process and the testing of new treatments. Given that bringing a new drug to market can take up to 15 years and be an extremely costly and risky exercise, many pharmaceutical companies are devoting more resources to finding new uses for existing products.
“Even though machine learning is used at all stages of drug discovery, it is of most use at the start, when there is a large amount of data to process and connect,” says Amanda Schierz, principal data scientist at AI platform DataRobot. “This is the best stage to predict any problems that could occur further down the line. Towards the end of the process, machine learning is used predominantly as a support tool for human decision-makers, enabling researchers to investigate alternative combinations and candidates quickly.”
She observes that the speed and productivity of automated machine learning align with the drug-discovery philosophy of ‘fail fast’. This helps to reduce the number of failures later in the process.
“With the public focusing on – and becoming more knowledgeable about – the drug discovery process in relation to Covid-19, it’s time for companies to rely more on machine learning,” Schierz says. “This enables them to innovate and fail fast, instead of putting all their resources into one metaphorical basket.”
