DeepMind and NHS hospital’s AI shows great promise in eye diseases
Google’s DeepMind unveiled early results from its AI-based research partnership with London’s Moorfields Eye Hospital yesterday, which promises to transform the management of sight-threatening eye disease.
The results, published in Nature Medicine, show that the firm’s AI system can quickly interpret eye scans from routine clinical practice with unprecedented accuracy, matching the skill of the most highly trained ophthalmologists.
The research found the AI correctly recommended how patients should be referred for treatment for over 50 sight-threatening eye diseases.
These are early results, but it shows that the system could handle the wide variety of patients found in routine clinical practice. In the long term, DeepMind says the technology will help the NHS and other healthcare systems worldwide diagnose and prioritise patients who need urgent treatment, and could ultimately save sight.
Moorfields says integrating the AI system into routine care would help improve care, reduce the burden on clinicians, and lower costs.
However, there is still a long way to go before the technology enters the NHS mainstream, with the next step being the start of clinical trials. Once these are underway, the commercial deal with DeepMind will allow Moorfields’ clinicians to use it free-of-charge across all 30 of its UK hospitals and community clinics, for an initial period of five years.
These clinics serve 300,000 patients a year and receive over 1,000 OCT scan referrals every day – potentially providing much improved accuracy and faster diagnosis.
The tech company says the original dataset held by Moorfields was suitable for clinical use, but not for machine learning research, which meant it had to invest significantly in “cleaning up, curating and labelling” the data, which it says is now one of the best AI-ready databases for eye research in the world.
DeepMind is having to tread very carefully in working with the NHS and patient data, as there is growing suspicion about tech and big data companies exploiting data for commercial gain.
In 2016, another NHS partner, The Royal Free Hospital in London, was found to have breached UK data law by handing over patient data to DeepMind without adequate patient consent.
That project – named Streams -is aimed at detecting acute kidney injury (AKI) continues – and also has the potential for improved patient care, but the partners have introduced more stringent data handling rules.
At Moorfields, the improved database is owned by the NHS Trust as a non-commercial public asset, and is already been used by hospital researchers for nine separate studies into a wide range of conditions, and with more planned. Moorfields is also able to use DeepMind’s trained AI model for their future non-commercial research efforts.
DeepMind co-founder Mustafa Suleyman
Mustafa Suleyman, co-founder & head of Applied AI at DeepMind wrote in a blog: “For all of us who have worked on this since we signed our agreement with Moorfields in 2016, this is a hugely exciting milestone, and another indication of what is possible when clinicians and technologists work together. We’ll continue to keep you updated as we make progress.”
Bayer buys into antiviral research for lung disease
Haplogen a Vienna-based biotech company and collaboration partner of Evotec, has entered into a multi-year drug discovery and development collaboration with Bayer to identify new therapeutics with applications in pulmonary diseases such as COPD.
Haplogen and Evotec have been collaborating since 2012, and have built a portfolio of pulmonary therapeutic programmes based on the drug discovery platforms and know-how of both companies.
The goal of the new Haplogen-Bayer research alliance is to develop new antiviral compounds addressing the high unmet medical need in reducing COPD exacerbations.
As part of the collaboration, Bayer receives an exclusive licence to worldwide rights to programmes developed by Haplogen and Evotec. Undisclosed upfront payment and potential milestone and royalty-based payments will also be paid by Haplogen to Evotec.
Roche’s Alecensa approved by Scottish Medicines Consortium
Patients in Scotland with untreated, advanced non-small-cell lung cancer (NSCLC) with ALK-positive tumours have a new treatment option available via NHS Scotland funding from today.
Alecensa is challenging Pfizer’s existing ALK+ lung cancer treatment, Xalkori, alongside another challenger, Novartis’ Zycadia.
The positive decision from the Scottish Medicines Consortium (SMC) was based primarily on the phase III ALEX trial, a head-to-head study in which Alecensa showed significantly superiority compared to Xalkori in reducing risk of disease progression or death.
As well as targeting the tumour in the lung, Alecensa also cuts the risk of tumours in the central nervous system (CNS), with the incidence of brain metastases high in this patient group.
The decision follows a NICE recommendation in this indication for patients in England and Wales in June.
The advice is conditional to commercial agreement with NHS Scotland and Roche.
Prior to this approval, Roche provided Alecensa via the UK-wide Early Access to Medicines Schemes (EAMS), which allows patients and doctors access to promising new unlicensed medicines that have a high unmet clinical need before official approval is obtained.
Novartis’ Zycadia was also recommended by NICE in January. The ruling is one of several new decisions the SMC announced yesterday. Click here for the full list.