New AI tools could speed up cancer diagnosis – and improve treatment

A new method of using artificial intelligence (AI) to predict cancer cell behaviour, potentially improving diagnosis and treatment, has been developed by the Hebrew University of Jerusalem.

The new diagnostic tool combines nano informatics and machine learning (ML) to classify cells based on the uptake of cell particles with different sizes. It could improve both the accuracy and speed with which clinicians can test cancer cell behaviour from patient biopsies, leading to the development of new clinical tests to monitor the progression of the disease and the effectiveness of the treatment.

Current diagnostic methods for cancer, such as imaging scans and tissue biopsies, are invasive, expensive and time-consuming, and can lead to delays in treatment. These approaches also offer limited insights into the disease’s behaviour at the cellular level. There is an urgent need for more effective and non-invasive diagnostic tools.

The research, which was carried out by doctoral student Yoel Goldstein, Professor Ofra Benny and Professor Tommy Kaplan, involved exposing cancer cells to particles of various sizes, each identified by a unique colour. They then looked at the precise quantity of particles consumed by each cell. Machine learning algorithms analysed these uptake patterns to predict critical cell behaviours, such as drug sensitivity and metastatic potential.

“Our method is novel in its ability to distinguish between cancer cells that appear identical, but behave differently at a biological level,” Goldstein said. “This precision is achieved through algorithmic analysis of how micro and nanoparticles are absorbed by cells. The ability to collect and analyse new types of data brings up new possibilities for the field, with the potential to revolutionise clinical treatment and diagnosis through the development of new tools.”

Benny said that the research paved the way for new types of clinical tests that could significantly impact patient care: “This discovery allows us to potentially use cells from patient biopsies to quickly predict disease progression or chemotherapy resistance and could also lead to innovative blood tests that assess the efficacy of targeted immunotherapy treatments.”

AI can reduce mistakes and improve patient safety

While AI is demonstrating its potential at the cutting edge of cancer diagnosis and treatment, it can also be used in a very different way – to automate administrative tasks routinely performed by clinicians, thus reducing workload and burnout.

In a recent talk at the Amazon Web Services summit in Washington, Dr Christine Tsien Silvers explained how generative artificial intelligence tools can autogenerate referrals, summarise research, draft clinical notes using ambient listening, explain medications to patients and check on patients who are admitted into the hospital. (Generative AI refers to the use of AI to create new content, such as text, images and music.)  By reducing clinician burnout, the AI tools would also help to reduce mistakes and improve patient safety.

Harvard Medical School, for example, is using generative AI to help interpret arterial blood gas (ABG) test results quickly. An ABG test, which requires a blood sample taken from an artery, measures oxygen and carbon gases in the blood. It can be used to evaluate conditions such as acute respiratory distress syndrome, sepsis, hypovolemic shock, an asthma attack, cardiac arrest, respiratory failure and heart failure. This test is often conducted in emergency situations, and the correct interpretation of its results can be crucial.

Dr Praveen Meka of Harvard Medical School explained to the conference how he and his colleagues applied generative AI to a database of ABG test results to increase the speed of interpretation, with an accuracy of 98%.