For a long time, people have been speculating about how the arrival of artificial intelligence would impact our lives. Pessimists believe that artificial intelligence will eventually spell the end of civilization; others with a more optimistic disposition believe that it will usher in a new era of precision and connectivity.
Speculations thrive when they are made about things that are yet to come. However, in the last year or so, artificial intelligence has begun to make an appearance in a pronounced manner: we have seen apps that can change facial expressions in real-time, apps that age people at will, and most recently, apps that can write clear-headed, realistic essays about almost any topic imaginable.
Read more about cholangiocarcinoma etiology
While artificial intelligence, in the minds of many, is a medium for entertainment/convenience, it has potential to do incalculable good, especially in the world of medicine.
The main advantage of artificial intelligence, indeed the crux of its merits, is that it can mimic human learning in a remarkable way. This is why no human chess master can ever beat a computer; it simply has the ability to take into account a near infinite amount of possibilities and make the best decisions accordingly. The ability of artificial intelligence to leverage large amounts of data to perform increasingly complex tasks is known as “machine learning.”
In Molecular Diversity, Gupta and colleagues wrote about how artificial intelligence is being used in drug discovery. Drug discovery is a long, tedious process because of the amount of research it requires. For example, there is the primary question of where the need for new drugs are. In order to understand the balance of supply vs demand, researchers need to have a deep understanding of the current landscape of drug availability and shortages.
Following that, scientists need to be able to identify relevant biomarkers to target and decide on the right molecular pathways to intervene. In addition, scientists need to figure out if there are any drug interactions that can lead to unwanted adverse effects.
All this generates a large amount of data. The beauty of artificial intelligence is that it is designed to process copious amounts of information; furthermore, it can “learn” from the data collected, helping clinicians spot patterns and find rhyme and reason to the information they are seeking to analyze.
“Though there are some unavoidable obstacles and tremendous amounts of work to be done to incorporate [artificial intelligence] tools in the drug discovery cycle, there is no doubt that in the near future [artificial intelligence] will bring revolutionary changes in the drug discovery and development process,” Gupta and colleagues wrote.
Assessing Cancer Outcomes
In Annals of Medicine, Zhou and colleagues highlighted another potential use of artificial intelligence — assessing intrahepatic cholangiocarcinoma outcomes and helping physicians decide on the best course of treatment.
Intrahepatic cholangiocarcinoma is the second most common primary liver cancer. The main challenge with this cancer is that patients are usually diagnosed late, owing to the ability of the cancer to avoid detection for years. As a result, radical surgery is not an option for the vast majority of patients.
Intrahepatic cholangiocarcinoma is usually managed in a multidisciplinary setting, given that it is usually treated with a number of modalities: chemotherapy, systemic therapy, and locoregional therapies.
“There is still no mature recommended regimen for the multidisciplinary treatment of [intrahepatic cholangiocarcinoma],” Zhou et al wrote. “Therefore, retrospective studies based on historical data are of great significance in clinical decision-making.”
They hence sought to build a predictive model that can assist physicians in deciding on the best course of treatment for their patients. The research team accessed data from the Surveillance, Epidemiology, and End Results (SEER) database and analyzed the clinical information of patients who were diagnosed histologically with intrahepatic cholangiocarcinoma. Among the clinical information extracted were sex, age, tumor size/stage, tumor invasion, peritoneal perforation, and cancer metastasis.
The researchers then input these data into prognostic models with the expectation that they can help clinicians arrive at optimal treatment decisions. Three prognostic models were created: the first was fed data from 4398 patients in the SEER database, the second was fed data from 504 patients from a single hospital and tested using 499 Asian patients in the SEER database, and the third received data from both datasets.
Read more about cholangiocarcinoma treatment
Given that the authors of the study were based in China, they opted to develop a web application based on the second model as it was most suitable for their Chinese patients. The verdict?
“Overall survival and mortality risk could be quickly estimated by inputting clinical parameters, TNM stage, and treatment options,” they wrote.
This study illustrates the great promise of artificial intelligence in clinical practice. As of now, most treatment algorithms are developed by clinicians and are updated periodically according to best available evidence. Artificial intelligence can greatly improve the accuracy of these algorithms and eliminate any risk of bias.
“Machine learning has been increasingly applied in clinical settings to assist physicians with better recommendations,” Zhou and colleagues wrote. “With the emergence of new treatment methods, current prediction models need to be updated to meet [the highest] clinical standards.”
Gupta R, Srivastava D, Sahu M, et al. Artificial intelligence to deep learning: machine intelligence approach for drug discovery. Mol Divers. Published online April 12, 2021. doi:10.1007/s11030-021-10217-3
Zhou SN, Jv DW, Meng XF, et al. Feasibility of machine learning-based modeling and prediction using multiple centers data to assess intrahepatic cholangiocarcinoma outcomes. Ann Med. Published online December 28, 2022. doi:10.1080/07853890.2022.2160008