In a recent review published in the journal Nature Review Cancer, Researchers argue why a fundamental understanding of the capabilities and limitations of artificial intelligence (AI) applications is becoming increasingly important in today's fight against cancer. They briefly introduce AI and related models (artificial neural networks (ANNs), deep learning, and large language models). [LLM]), and highlight the advances in the field and their application in cancer research, and the challenges facing the ubiquitous adoption of AI technology in ongoing studies.
This review aims to serve as a practical guideline for the adoption of AI into mainstream cancer research, primarily targeted at non-computationally inclined cancer biologists. It provides several examples of how technology can accelerate research progress and identify patterns invisible to the naked human eye.
Review Article: A Guide to Artificial Intelligence for Cancer Researchers Image credit: Springsky/Shutterstock
What is AI, and why is it important in cancer research?
Artificial intelligence (AI) is an umbrella term for many technologies and applications that attempt to simulate human intelligence and data processing using high-precision machine algorithms. Despite its broad introduction during a conference in 1956 (Dartmouth College), AI has remained a theoretical principle-based system for most of its existence, with what are now 'symbolic AI' and 'classical machine learning' dominating the field. But dominated. As recently as the last 15 years.
Unprecedented advances in simple artificial neural networks (ANNs), back-propagation algorithms, and, more recently, deep neural networks (DNNs) and large language models (LLMs) have taken the field beyond its theoretical roots and research and It has seen its widespread adoption in industrial applications. . Recent releases to the public of LLM- and deep learning-driven applications such as Gemini AI and ChatGPT have further accelerated the development of AI, medical research on these technologies for rapid diagnosis, drug discovery and data analysis. Dependent.
“…we assume that any cancer researcher today needs to achieve a certain level of AI literacy. Today, it is important to understand, interpret and critically evaluate AI output. A deep understanding of AI and developing their own AI-based software tools Today, AI is commoditized, meaning it is no longer a specialized resource but a widely accessible tool. “Can easily be used by cancer researchers.”
Cancer research is no different, with AI-based applications increasingly used in cellular and molecular image processing, histopathology research, and radiology. LLMs, in particular, are increasingly being used to collect and analyze clinical data, greatly improving the rate at which data can be processed and the subtle patterns and trends within the data. Helps to identify which are often missed during manual human search.
About the review
The present review attempts to convince cancer researchers, especially those who are not computationally inclined, of the potential of AI and related technologies in advancing our understanding of disease and how to combat it. About the benefits. The authors cite more than 170 clinical and computational publications while tracing the evolution of AI from its theoretical roots nearly 70 years ago to the more familiar practical applications we see today.
After that, he narrowed his introduction to AI to focus on current and potential applications of the technologies in cancer research and therapy. They highlighted the easily accessible 'off the shelf' software available to every cancer researcher, regardless of what they need to keep in mind when interpreting results from some of these platforms.
Understanding Deep Learning
In addition, the researchers introduce the theoretical framework governing classical machine learning algorithms and how they have evolved into today's deep learning technologies. They distinguish between different types of deep learning (supervised, unsupervised, and reinforcement) and their current applications in cancer research. A key element of this section is the automation provided by the deep learning platform and the substantial time savings (productivity) it can provide over traditional analytical methods, especially during large clinical trials.
Biomedical Image Analysis
This section highlights the use of AI in image detection, identification and classification. It traces the evolution of its clinical application from the classical machine learning methods of the late 1990s and early 2000s to the much more complex algorithms of today. The former was used to detect and sequence microscopy images, while the current has advanced enough to use biomarkers to diagnose the type and severity of cancer.
“Many image analysis tasks in biological research are traditionally done manually, however this is not only inefficient and error-prone, but can also make experiments impractical if thousands of output images are to be analyzed. , analysis can be made more objective, reliable and faster, for example, in the context of detecting cells in phase contrast microscopy, deep learning can quickly and reliably detect individual cells and Such analyzes are being performed on a large scale. For example, commercial platforms such as the Incucyte AI Cell Health Analysis Software Module (Sartorius AG) are used.”
This section introduces standard commercially available deep learning tools for implementing AI in histopathology and computational pathology assessments while also suggesting that some custom-built deep learning tools are not as complex to code as That those of us non-computationally inclined might believe. This section further lists some of the challenges faced by the adoption of AI in biomedical image analysis, the most important of which is 'descriptiveness' – given the relative modernity of the technology, AI tools Some of the patterns identified by cannot (yet) be explained. However, recent changes in AI algorithms and the use of clinical trials to validate some of these obscure paradigms are helping to overcome these challenges.
Drug discovery
Large transformer models, a new subclass of AI technologies, are making significant progress in the field of cancer drug discovery. In contrast to traditional applications, these models can predict the binding and efficacy abilities of candidate therapeutics in the active regions of a patient's protein, thereby reducing the uncertainty involved in current and future clinical trials.
Results
The most important AI challenge in cancer research today is mining real-world data (RWD), including EHRs, tumor samples, and medical images. Unlike clinical trial data, which typically follow well-defined procedures, RWD is typically random in both its collection mode and documentation, adding to the complexity of its analysis. Even if there is an increase. Despite the challenges, the unprecedented growth and adoption of AI paints an exciting future for oncology, and basic literacy of its warnings is increasingly becoming a necessity, not just for budding cancer biologists. .