Hope KellnerArtificial intelligence (AI) is a popular and evolving technology that is often defined as computer programs or algorithms trained to “think like” or replicate human intelligence. The applications of AI are endless. With cancer continuing to be a significant public health issue, AI offers the opportunity for improving patient outcomes and detection accuracy.
There are many different types of screening for cancer detection involving human analysis such as looking for abnormal molecules in the blood, genetic testing, examining specific tissues (i.e. pap smears), and cancer imaging (i.e. breast cancer mammograms). However, only 14% of cancers are detected through the preventative screenings, which highlights the need for the development of additional screenings to detect more cancer types as well as increase the efficacy of current testing. AI can assist with a part of the problem– the human biases and misinterpretations introduced during the screen analysis– by reducing false positives and false negatives. A study developed an AI system outperformed human radiologists in breast cancer detection. The University of Pittsburgh also developed an AI algorithm that was able to detect prostate cancer with high sensitivity and specificity. Although, it is important to note that the AI system was trained on patient cohorts across only two countries, which leaves room to add more diverse training. These are just a few examples of the progress made utilizing AI in cancer detection and interpretation.
AI can also help with providing more detailed predictions on a patient’s prognosis, or the likely outcome of treatment for the specific cancer. A number of factors are involved when predicting a prognosis: type of cancer, size and spread of cancer, genetic contributions, age, health, treatment route, etc. Doctors consider historical trends to give a prediction of a patient’s prognosis, and although a highly-educated estimate can be given, it is often a lengthy and time-consuming process to take all the factors into consideration. This is where AI can come in and analyze the patient’s specific situation and predict the likelihood of the disease progression and treatment outcome. The emergence of the constantly-evolving AI and its integration into healthcare greatly improves upon personalized medicine capabilities.
However, there are a lot of limitations to AI that need to be taken into consideration. Although AI algorithms are exceptionally helpful for cancer detection and prediction, they are not meant to replace clinicians or other doctors. There are also some ethical concerns with AI involvement including patient data privacy, competition between institutes leading to decrease in data quality, and potential medical bias if the AI is not trained properly with diverse representation of the population. Overall, AI has proven to be an undeniable asset to medical teams in cancer diagnosis and prevention, but it’s critical to understand its limitations and establish medical AI ethics and protocols.
Peer Editor: Grace Stroman