(Original post at Tremr)
Research from the University of Adelaide, recently published in the Naturejournal Scientific Reports , has revealed that artificial intelligence can be used to predict the lifespan of patients by examining images of their organs.
Using images of 48 patients’ chests, the computer predicted with 69% accuracy which patients would die within 5 years. The research suggests that AI may be used in the future to diagnose serious illnesses in their early stages, tailoring treatments to individuals and potentially saving lives.
The study was conducted by researchers from Adelaide’s School of Public Health and School of Computer Science, alongside collaborators from Australia and elsewhere in the world. Researchers will now proceed to analyse tens of thousands of patient images to continue their investigation.
The University of Adelaide’s study is not the only way in which artificial intelligence is influencing modern medicine; healthcare may represent the most promising avenue for AI research in the coming years.
From diagnosis and treatment support to increased hospital efficiency and patient homecare, medical institutions are beginning to embrace the possibilities of artificial intelligence.
Famous for winning Jeopardy! in 2011, IBM’s Watson computer system has been applied in clinical decision support in hospitals, allowing physicians to describe a patient’s symptoms, medical history and other significant factors to provide a personalised list of recommendations for treatment. IBM Watson for Oncology, designed to offer options for cancer-care, was launched in 2016.
By increasing hospital efficiency, AI has the potential to greatly improve the quality of care patients receive. Dutch company Zorgprisma Publiek, for example, analyses medical invoices through IBM’s Watson system, using the data to identify where treatment mistakes are repeatedly made. Startup Enlitic also uses stores of medical data coupled with its deep learning techniques to draw conclusions and improve patient diagnostics.
Google’s DeepMind programme has been used by University College London Hospital to advance head and neck radiotherapy treatments; the AI allows doctors to more quickly ‘segment’ the cancerous cells, i.e. differentiate between healthy and unhealthy tissues, a process which can be lengthy for head and neck treatments.
Elsewhere in the NHS, Royal Free NHS Foundation Trust hospitals in London are using the alert app Streams, a system that utilises DeepMind to instantly update doctors about their patients and search digitised records more efficiently than paper files. Other apps such as Babylon Health, Molly by Sense.ly, and AiCure all aim to utilise AI to facilitate care for patients.
The future of AI in healthcare is promising, but not without potential ethical dilemmas, and as such the implementation of plans for artificial intelligence in healthcare must be carefully thought out.
The data-sharing agreement between the NHS and DeepMind, for example, has been criticised as invasive by many; a report published by the New Statesmanin 2016 revealed that DeepMind has access to 1.6 million patients’ medical records, though the company is obliged to delete the data once the agreement expires in September 2017.
National Data Guardian Dame Fiona Caldicott also deemed the data-sharing agreement to be “on an inappropriate legal basis” in a leaked letter.
Director-General of the World Health Organisation Margaret Chan has also highlighted that whilst artificial intelligence can be a force for good in healthcare, the expense of AI makes it currently nonviable for many developing countries.
Speaking at the UN AI for Good Global Summit, Chan remarked, “enthusiasms for smart machines reflect the perspectives of well-resourced companies and wealthy countries.” She added, “what good does it do to get early diagnoses of skin or breast cancer if a country does not provide the opportunity for treatment or if the price of medicines are not affordable?”
These issues do not invalidate the possibilities of narrow AIs like Watson or DeepMind, but, like any new research or treatment method, artificial intelligence must be assessed and implemented properly by medical professionals. Utilised properly, artificial intelligence could completely revolutionise modern medicine.