Healthcare AI in 2020: the ones to watch

Chloe Kent 22 January 2020 (Last Updated January 21st, 2020 12:13)

2019 was a big year for artificial intelligence in healthcare, and 2020 is set to be no different. Chloe Kent rounds up the projects to keep an eye on over the next 12 months.

Healthcare AI in 2020: the ones to watch
Many AI enterprises will make headlines in 2020. Credit: Shutterstock

Artificial intelligence (AI) is beginning to touch medicine in many different ways, from making diagnoses to triaging patients. It has proven problematic at times – just think of the algorithm which favoured healthier white patients over less healthy black patients – but smooth out the kinks in the coding and its applications are potentially revolutionary.

Many AI enterprises will make headlines in 2020, but at Verdict Medical Devices we’ve picked out the three we think deserve a little extra attention.

Google’s Project Nightingale: ever so slightly suspicious

Google didn’t exactly come out of 2019 smelling of roses when it comes to patient data privacy. Alongside the clamour in the UK following the news that the tech conglomerate had finally fully absorbed DeepMind, an AI company criticised for its historic handling of sensitive NHS data, its US-based cloud computing deal Project Nightingale found itself under regulatory fire.

The partnership between Google and Ascension Health is set to integrate Ascension’s numerous health data streams into the cloud. They’re then set to be processed through an AI and give healthcare providers new insights and suggestions for patient care. However, the US Office for Civil Rights in the Department of Health and Human Services initiated an investigation into the project after it learned that this information was not anonymised before being accessed by at least 150 Google staff.

Shortly afterwards, an anonymous whistle-blower uploaded a now-deleted video to Daily Motion showing a document dump of hundreds of images of confidential files relating to Project Nightingale. The video detailed the transfer of millions of patients’ personal records to Google and suggested that in the future Google might be able to sell or share the data with third parties or create patient profiles against which it would advertise healthcare products.

US Senators Elizabeth Warren and Richard Blumenthal wrote to Google expressing their concerns and imploring the tech giant to answer a series of 17 questions about how the data was collected and what the company planned to do with it. Google defended its practices, and denied that it would use any personal health information to target adverts to customers.

All things considered, this makes Project Nightingale an AI enterprise to keep an eye on over the coming year. Perhaps the project is well-intentioned and will make waves in patient care. Or maybe the ads in your Facebook sidebar will start to feel far too close to home.

Abtrace battles against antimicrobial resistance

Antimicrobial resistance (AMR) is one of the greatest ongoing threats to global health, estimated to cause 25,000 deaths and 2.5 million extra hospital stays per year in Europe alone. Overuse and misuse of antibiotic drugs leads to different bacteria becoming resistant to even the most powerful antibiotics available, making the diseases they cause difficult or even impossible to treat.

Leading the fight against antimicrobial resistance in the tech world is Abtrace, an AI platform designed to help clinicians prescribe the most appropriate antibiotic for each individual patient they see in their practice. When around 30% of antibiotic prescriptions are inappropriate, this couldn’t come at a more vital time.

The platform makes its recommendations through a process known as natural language processing (NLP), where a patient’s healthcare notes are processed through Abtrace’s augmented decision-making tool. The algorithm contains millions of data points on dosages and success rates in treating different infections with different antibiotics, against which the patients’ notes are compared. In seconds, it presents a recommendation for whether or not an antibiotic should be prescribed, and which antibiotic would be appropriate if so.

By helping to guide doctors to the right decision, the risk of antimicrobial resistance developing is drastically reduced. Plus, it has the added bonus of helping less people waste time and money on inappropriate medication regimens.

Abtrace is a European Institute of Technology Health (EIT Health) Wild Card Project, and will receive up to €2m from the organisation to help commercialise the product.

Pexxi genetic testing aims to decode contraception

Many women who choose to use hormonal contraceptives, such as the pill, implant or ring, have to navigate several different options through trial-and-error until they are able to find a medication that works for them.

There are currently 28 brands of contraceptive pill available in the UK, and even within the subcategories of ‘combined’ progesterone and oestrogen and ‘mini’ progesterone-only pills, their hormonal and chemical makeups can vary greatly. The side effects associated with hormonal contraceptives, such as acne, weight gain, anxiety and depression, can have a huge effect on a person’s life, and it can take months or even years before a woman is able to settle on one with no or only minimal adverse effects.

Healthtech start-up Pexxi is now aiming to end the contraceptive roulette wheel through AI-powered genetic testing. Users give Pexxi a spit sample, which contains enough information about where their progesterone and oestrogen levels naturally sit as well as whether they have any genetic predispositions to the potential side-effects of one type of pill over another. The results are mapped on to the index of hormonal contraceptives the AI knows about, which is constantly updated with the latest research, to find the best match for the individual. Pexxi then provides a list of the contraceptive pills the person in question is most likely to tolerate, with plans to eventually expand to include the implant and ring as well.

Pexxi is currently in beta-testing stages, with a plan to eventually reach customers through a 23andMe-style model where they’ll pay a fee to use its services.