Whether you are a medical researcher or practicing health professional, it is a daily battle to fight through the tonnes of information that seem to pour in from all directions. Academic articles, preprints, clinical trials and media reports can all add to the confusion where they should help optimize decision-making.
With the need to ensure the wealth of knowledge out there is fully utilized, AI is increasingly becoming the tool of choice to solve this conundrum. Recognising this need, publishers and information providers are pushing AI’s capabilities to streamline workflows and optimize researchers’ time and effectiveness. For example, a number of organizations are using AI to categorize data in different ways, whether it be in terms of where articles fit in the UN Sustainable Development Goals (SDGs) framework, or in terms of the geography of where research has been undertaken. Indeed, Dimensions is one platform in which AI has been used to categorise documents into previously defined topic areas in an automated way for many years now.
There is perhaps no other area where the impact of research is being felt more keenly than in cancer research, where according to the National Cancer Institute in the US it will spend nearly $7bn in 2022. On a smaller scale, AI should be able to be put to good use extracting the maximum value of these dollars, and one example of this has been explored in the recent article Application of artificial intelligence to overcome clinical information overload in urological cancer by Arnulf Strenzl et al in BJUI International.
In the article, the authors look at how the use of AI can help researchers – specifically in the field of uro-oncology – set up more sophisticated queries than traditional methods. The benefits of this approach can provide what the authors call a “synthesis of raw data and complex outputs into more actionable and personalized results”.
The study examined a number of tools, including Dimensions from Digital Science, and the use of AI within them. The advantages that AI offers beyond traditional search methods are clear when it comes to looking at evidential support for treatment choices or sequencing decisions. A straight search in relation to a specific disease in the PubMed database, for example, will yield thousands of results, with much more work required to filter those results on article or journal performance before the most relevant sources appear, and even then there may be dozens or even hundreds of possibilities.
What AI can produce – and in the article Pfizer’s dashboard on prostate cancer built with Dimensions is used as a prime example – is the ability to quickly cross-reference dozens of treatments and identify where studies have taken place. The article does also highlight some limitations in the dashboard – we are still some way from simply feeding documents into a machine with decisions made for us, and so research still requires an element of expert interpretation. However, with constant improvements being made using technological advances and feedback, future enhancements are never very far away.
For researchers and practitioners in this space, the article concludes that the Dimensions platform, “demonstrated how AI can assist urologists and oncologists to readily obtain focussed information on a specific topic of interest from rich datasets”, for example by gaining insights into published articles and clinical trials on specific combinations or sequences of therapeutic agents. It further states that AI, “exhibits the potential to support more tailored, accurate, and reliable clinical decision-making from the multiple sources of information available”. While AI may never get close to delivering perfect answers to researchers’ questions every time, it can certainly succeed in delivering better answers more quickly than ever before.