5/07/2021 |

Health and Intelligent Machines

Lino Mari | Head of Technology, Healthware International


The technological revolution now permeates every context of our lives and the impact is disruptive. Often the activation of new technologies depends primarily on the ability and attitude they are received with.

In the health sector, we are all familiar with technological revolution and speed at which science evolves, but today we are facing a new challenge that has entered the medical field with gusto: Artificial Intelligence.

This technology is already present in many areas of medicine, for example, in care or to support diagnoses. Artificial Intelligence models for decision support have been developed in many areas of medicine and their near future impact and potential are very clear.

The nature of data in healthcare – a high volume of information to be analyzed – finds the application of artificial intelligence and machine learning algorithms as inevitable support.

Medical image classification AI applications have demonstrated such excellent performance that in some cases, it has surpassed medical experts. But these systems will only provide a real clinical benefit if the physicians using them are confident in the results they provide.

To better understand the mechanisms of adoption among physicians, a study was proposed that basically allowed them to ask for a second opinion, either from other professionals or via AI system support. The study revealed some skepticism toward the advice provided by AI even when it was more accurate than that provided by professionals.

More information about the study can be found here: https://www.nature.com/articles/s41746-021-00385-9).

Diagnostic accuracy

The study demonstrated the effect of advice accuracy and source on diagnostic accuracy for task experts (radiologists) and non-experts (IM/EM physicians). Graph (a) compares diagnostic and advice accuracy, demonstrating that both physician groups perform better when they receive accurate advice.

Table (b) compares the diagnostic accuracy of AI and humans, demonstrating that neither group of physicians had a significant difference in diagnostic accuracy based on the advice source. There is no significant correlation between advice accuracy and advice source.

This means that in clinical practice, the technology is ready to be used successfully, but the road to adoption is still long.

Artificial Intelligence and Drug Discovery

AI is already very advanced in the pharmaceutical sector, so much so in the discovery of new drugs and vaccines that it is part of a generalized digitalization process.

Today, pharma companies are investing in startups that are implementing new AI-based platforms to take drug discovery and development to a new level. There are already important and valuable applications in the areas of compound studies, pre-clinical studies, and clinical trials.

Compounding studies improve and accelerate the time to drug selection and validation using AI-based virtual systems and access to huge compound databases. Pre-clinical studies replace laboratory testing to assess the safety and efficacy of a new drug. By using deep learning algorithms, some pharmacological properties can be simulated and predicted.

AI also helps improve portions of the drug manufacturing process that involve predictive maintenance, quality control, and triggering processes that do not require human interaction, eliminating many errors.

Example: Cyclica

Cyclica, which was founded in 2013 and specializes in AI and data-driven drug discovery, launched its deep learning platform that allows for proteome-wide evaluation to determine complex drug poly-pharmacology in real time, meaning it predicts the on-target and off-target interactions these compounds can have in vivo.

Traditional approaches to drug screening tend to focus on a single protein target paradigm (potency), but compounds will hit multiple targets in the body. This can lead to off-target effects such as efficacy and toxicity issues that could cause a drug to fail in early preclinical or clinical testing.

AI in Neurodegenerative Disease Treatment

Big data produced by research over the years is now being used to discover patterns and discovery acceleration models. For example, the University of Cambridge has used machine learning algorithms to decipher diseases such as Alzheimer’s.

In this field, machine learning systems applied to research represent a real change. Thanks to these algorithms, new drugs and treatments can be developed to fight symptoms or prevent other disease effects.

Example: Imeka

Imeka is a company that is working to combine diffusion imaging and AI to map white matter integrity and get insight on neuroinflammation, demyelination, and axonal loss. Non-invasive technology minimizes risks for pharma and biotech clients in the development of treatments for neurodegenerative diseases.

A noninvasive MRI based biomarker technology was used to analyze free water as a marker of neuroinflammation in patients. Although preliminary, the observed decreased neuroinflammation results are encouraging.

This technology makes it possible to examine neuroinflammation in a way that has not been previously available and showed a robust reduction in neuroinflammation in an area of the brain important to Alzheimer’s disease in a small number of patients.

In conclusion, leveraging Artificial Intelligence for clinical decision support, risk assessment, and early warning are the most promising areas of development. The new generation of tools and a better approach to adoption by clinicians will make for more efficient delivery of care and usher in a new era of clinical quality and exciting patient care advances.

Let’s talk about what truly matters and how we can help you in building an AI-based app for your business needs.

  • Artificial Intelligence
  • Big Data
  • Digital health
  • Digital Health Transformation
  • Technology
Lino Mari

Lino works with international teams and supports clients in identifying the right fit of Software Architecture and Technology in order to develop corporate communications, campaigns, service portals and mobile enabled websites. Lino loves R&D, Agile (Scrum), Big Data and Open Data.