Healthcare: 3 key activities that AI will transform

Published on 22 April 2020 Read 25 min

AI is a recurring topic in the media. Claims have been made on the game changing features of AI and on the fact that all sectors will eventually be affected. But what is the situation regarding healthcare? Innovation in healthcare has always been a sensitive topic especially when it involves a greater role of the machine. At Alcimed, we looked at 3 activities in healthcare that AI will empower.

1. AI for optimized drug development

When following a traditional drug discovery approach, at least 10 years and around $2.5 billion are needed to go from conception  to commercialisation. Drug development is indeed a major challenge for pharma companies: it takes a lot of time, money and resources to bring a new drug to market. However due to fierce competition among pharma companies, they all invest tremendous resources to benefit from the first-mover advantage. Artificial intelligence can alleviate numbers of these pain points at different steps of the drug development process. AI helps namely with:

Identifying drug target and suitable molecules from data libraries

Drug design software are trained to spot the features and the mechanisms of action of known drugs and can then create new molecules from scratch meeting the requirements.

Suggest chemical modifications

The great computing power enables the machines to consider different drug designs at the same time. AI conducts the same experiments a researcher would but through computer simulation saving the researcher time for more complex tasks.

Identifying the best candidates for clinical trials

AI can go through plenty of data in seconds and select the candidates with the best fit for a trial in a very rigorous manner.

These use cases do not come from a futuristic movie but are actually occurring right now! The first drug designed by an AI has entered phase I of clinical trial early 2020. This is the result of a joint venture between the Japanese Sumitomo Dainippon Pharma and the British Exscientia.  DSP-1181 is a long-lasting, potent serotonin 5-HT1 receptor agonist and is intended for the treatment of obsessive-compulsive disorder. Usually, 5 years are needed before a new drug enters clinical trials. AI has shortened this to 12 months.

This drug is the first of many. According to experts, automation is the future. We will evolve from an augmented drug design process – in which all the decisions are made  in the end by the researcher – towards an autonomous drug design step in which the choice of the next compound to make will be made autonomously by the machine.

2. AI in healthcare: the doctor’s ally for diagnosis

Examining countless medical images is very time-consuming for pathologists and radiologists, once again when their time and expertise could be of better use elsewhere. And AI can make diagnosis cheaper and more accessible throughout the world. AI is particularly efficient when the data the doctor examines is already digitized, and this is the case for all the following imaging:

– CT scan
– Electrocardiogram
– X-ray
– Eye and skin images

With training, AI can suggest a diagnosis in a split  second, such as lung cancer or stroke, through CT scans or point out indicators of diabetic retinopathy based on eye images. The goal is not that AI alone establishes a diagnosis but rather to draw the doctor’s attention to what has been detected, allowing  the expert to focus on the interpretation of the signals. The idea for diagnostics companies could be to incorporate AI into their products and upgrade their offer.

Some AI of image analysis have already proven their efficiency, sometimes even surpassing humans. Mammoscreen, a product developed by the French start-up Therapixel, analyses mammography and is able to detect calcifications and masses and assess if they are benign or malignant. Two years ago, the start-up claimed to detect 75% of breast cancers compared to the 70% detection rate of radiologists [1]. AI is a relevant ally especially in terms of early breast screening.

Instead of having 2 specialists’ opinion on one single image like current regulations require, we can imagine a future with one first analysis established by an AI followed by a doctor’s approval or not. Further away in the future, diagnosis will be a much more global experience where data combined from multiples sources (CT, MRI, genomics and proteomics, patient data…) will be brought together for the AI to make a suggestion that will then be considered by the doctor.

3. AI in healthcare: a personalised prevention tool

AI has an important role to play in preventive care, namely with the use of predictive analytics. More than ever, personal data is being collected by various players and is being used to create a more personalised experience. This data collection, mainly through the use of health trackers, will best serve patients with the following:

– Risk of developing a certain pathology (risk factors)
– Earlier detection
– Follow-up of patients and anticipate possible complication due to treatment or surgery

RiskCardio falls in the first category. This machine developed by researchers at MIT’s computer science and artificial intelligence laboratory (CSAIL) determines the risk of cardiovascular death in already at-risk patients. RiskCardio classifies patients into risk categories after monitoring their ECG signal. This personalized approach enables doctors to adapt the treatment to the patient’s risk level which significantly increases  the chances for successful care.

Earlier detection will be empowered by IoT (internet of things) and even now IoMT (internet of medical things), both used to obtain a clear picture of everyone’s lifestyle. Data from various sources can be gathered (people’s smartwatch, smartphone apps, virtual assistant…) and met in order to know more about people’s manner of living. This will enable people to be alerted about possible red flags and foster earlier detection of pathology. AI may then decide which information is relevant to send to doctors, especially considering people’s medical background. With a clear insight on the patient’s way of life, the doctors – with the help of AI to analyse the data and alert them – can provide better guidance and support a healthier lifestyle.

Finally, AI and predictive analytics also improve patients’ follow-upin terms of comfort and medical supervision. . Patients may for instance be sent home from hospital earlier than in the past because tools will be there to monitor them and anticipate potential complication. We are evolving towards earlier detection/ diagnosis, more comfort for patients and better follow-up. Medical device companies could position themselves as aggregators of this data, collect it and process it through AI to provide relevant insight to patients and doctors.


The integration of AI in these three fields are not so much threats to pharmaceutical, diagnostic and medical devices companies as they are opportunities. These firms need to embrace AI and the innovations coming along. Alcimed has a deep interest in these topics and is ready to explore them with their clients!

About the author

Roxane, Consultant in Alcimed’s Healthcare team in Germany

[1] DMIST study (Digital Mammographic Imaging Screening Trial)

Do you have an exploration project?
Our explorers are ready to discuss it with you

Contact our explorers >


You have a project?

    Tell us about your uncharted territory

    You have a project and want to discuss it with our explorers, write us!

    One of our explorers will contact you shortly.