Can AI make healthcare more human?

Chloe Kent 20 February 2020 (Last Updated February 11th, 2020 18:09)

AI has the potential to transform the daily job of a doctor, from streamlining administrative tasks to providing fast and accurate diagnoses of conditions which can mystify humans. By automating these tasks, could AI make medicine more people-centric?

Can AI make healthcare more human?
When doctors spend less time filing paperwork and more time working directly with patients, they’re able to form a stronger connection with the people in their care. Credit: Shutterstock

In his book Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again, Dr Eric Topol explores how doctor-patient relationships have begun to erode. Physicians, Topol maintains, are overworked and overwhelmed and have become prone to making medical errors and misdiagnoses. He proposes that artificial intelligence (AI) has the potential to free physicians from the logistical tasks that interfere with their ability to connect to patients.

Contrary to concerns that tech is actually distancing patients from healthcare – commentators have expressed concerns that new-fangled interventions inadvertently isolate older patients, that people do not want to accept diagnoses from AI, that prejudices within an algorithm can prevent care from reaching the right people – Topol and others like him believe AI can actually help make healthcare more human.

Anyone who has kept their finger on the pulse of medical trends in recent years will be well aware that AI has the power to revolutionise the way healthcare is delivered across the board. The area where it has the potential to be the most transformative is also, arguably, the least exciting – administration.

Keeping the human touch

Draper & Dash’s software, which has been deployed more than 20 hospitals across the US, UK and Australia, uses predictive analytics to automate the clinical system that captures all of the patient transactions and data processed through the hospital. Meanwhile, Nuance Communications has built an automated digital dictation platform for doctors, which aims to help speed up record-keeping for physicians by transcribing spoken patient notes for them in real time. Digital triage assessment service Doctorlink allows a patient to undergo an automated consultation, after which an algorithm can direct them to a pharmacist, general practitioner or emergency care as appropriate.

By reducing the administrative burden on clinicians, AI software can enable them to spend more time working on the diagnosis and treatment of their patients.

“One of the things that we keep on hearing everywhere is that there are a lot of inefficiencies in pretty much every healthcare system,” says Doctorlink chief technology officer Alfonso Ferrandez. “A lot of that has to do with the administrative burden and the number of steps that even a patient has to go through in order to get the final outcome, be it a diagnosis or treatment.”

If a patient needs to see their GP, they can book an appointment straight through Doctorlink, which is then automatically registered in the practice’s clinical system with the symptom assessment summary included in advance of the appointment.

Dr Ravi Tomar, who uses Doctorlink in his clinical practice, says: “This system gives patients immediate access to a plan of care for their symptoms without removing the human touch, while freeing up clinician time to focus on complex, multi-morbid patients who will most benefit from a face-to-face appointment.”

It is this relief of burden which cuts to the heart of how AI can make healthcare more human. When doctors spend less time filing paperwork and more time working directly with patients, they’re able to form a stronger connection with the people in their care.

Blumberg Capital founder and managing director David Blumberg says: “In hospitals and health systems, care providers spend far too much time on administrative and repetitive tasks that can be managed by technology. AI can make healthcare more human by taking over these ‘drudge work’ tasks, allowing time for healthcare professionals to focus on the human activities.”

Diagnosis and treatment

AI can also make a clinician’s job easier when it comes to diagnosis and treatment. A mobile app developed by an Australian tech firm was able to record the sound of a patient coughing and use it to diagnose the cause of their illness, picking up on nuances easily missed by the human ear, and the possibilities for AI diagnoses through imaging pathways are only just beginning to be realised.

Outside of the hospital, AI can help constantly monitor a patient’s symptoms and recommend changes in their treatment. Researchers have used machine learning techniques to track the tremors of Parkinson’s patients, meaning their symptoms can be monitored on a more granular level and more targeted and appropriate treatment given.

Interventions like these also serve to make healthcare more human, by enhancing the quality of care a physician can provide. Why subject a patient with Parkinson’s to weeks of subjective self-assessment, which may not be especially accurate, when and AI can track and provide an objective analysis of their condition?

Of course, this doesn’t mean an algorithm can ever replace a trained, human physician.

Base2 Solutions senior systems engineer Dr Paul Kostek says: “While AI will be able to assess large amount of data, and provide doctors with options in terms of the patient condition, doctors and other members of the medical profession will still need to assess the data versus the symptoms the patient is showing to ensure the correct treatment is provided.”

Healthcare AI is still very much in its infancy, and it has a fair few kinks to work out first before it becomes standard practice. But a survey by Blumberg found that 81% of consumers believe AI will improve patient care, and with massive potential to improve administration, diagnosis and treatment, it’s hard to say they’re wrong.

Kostek says: “Human interaction is still a key element of medical care and, even with advances in AI, I don’t expect that to change.”