Use of AI in pulmonary oncology in combination with digital pathology, molecular testing, and electronic health record (EHR) data, can greatly aid in selection of an appropriate treatment drug and customization of radiotherapy dose. Based on data obtained from various studies, AI in pulmonary oncology and personalized medicine is also creating a breakthrough in lung cancer prediction, early detection, cancer characterization, prediction of patient survival, prediction of treatment response, customization of radiation dose, as well as identification of those patient who will have the greatest benefits from undergoing ICI therapy. Thus, AI has great potential regarding improving diagnosis, treatment, and outcomes for lung cancer patients.
Due to the perceived benefits such as reduction in post-op length of stay, fewer post-op complications, decreased bleeding, reductions in post-op pain, faster recovery time, and improved surgical outcomes, robotic-assisted surgeries have been widely popular in various surgical fields for many years (Mafi, Mafi, & Malahias, 2016; Wei, Eldaif, & Cerfolio, 2016). While researching papers for my capstone project, I was expecting to find more material regarding the use of AI and robotic-assisted surgeries in pulmonary oncology, however, that was not the case.
Robotic-assisted NSCLC surgeries have been performed in the past and da Vinci® Surgical Robotic System, which is the only robotic system approved by the US Food and Drug Administration for lung surgery, was used (Wei et al., 2016). Robotic arms were manipulated with the help of AI, and the outcomes of the performed surgeries were promising. Unfortunately, very complex setup time, high costs, absence of haptic/tactile feedback, and the need for specialized equipment and training were some of the limitations mentioned as the main reasons for surgeons not pursuing robotic surgery in lung oncology (Wei et al., 2016). In addition, the mediastinal area has a lot of lymph nodes and it can be quite difficult to resect lung cancer without removing lymph nodes, so some modifications in technology are needed in order to make this surgical procedure fully success (Belsley, 2011).
Robotic surgery is being used in numerous surgical fields such as pediatrics, urology, cardiovascular surgery, gynecology, otolaryngology, general surgery and orthopedics. It has resulted in a reduction in length of stay, post-operative complication and scarring. However, for achieving optimal outcomes, further development in improving the sensory feedback and reducing the lag time during the transmission of long-range telesurgery is required
My findings could be beneficial in radiology, pathology, histology, genetics and oncogenomics, and oncology. Radiologists could use ML in medical imaging so they can read and interpret images faster or compare images to see how patients responds to treatment. In addition, my findings could be also utilized in genetics, pathology, and oncogenomics. In this case, ML could be used to detect any possible mutations, which would make it easier for oncologists to determine the best treatment for their patients. If the lung cancer patient has ADC and either EGFR or ALK biomarker, targeted treatments could be applied.
Histologists, pathologists, and oncologists could not only use ML to extract certain tissue features from the slides and compare healthy versus cancerous tissues, but also compare between the two main NSCLC subtypes: SCC and ADC, and based on that, oncologists could personalize treatment plans for their patients. Based on the existing mutation and subtype of NSCLC, oncologists could also use ML to predict patient survival outcome. Moreover, DL could be used by pathologists to detect the presence of cancerous cells by identifying predictive patterns in images obtained through WSI.
Radiologists, pathologists, and oncologists could also apply DL to predict how patients who are treated with chemotherapy and radiation respond to those treatments, and if the dose needs to be modified, they could do so while the patient is undergoing treatment. Another way that my findings could be used is by radiologists, where DL could be valuable in customizing radiotherapy dose for the patient based on the lung imaging and patient’s features recorded in EHR. Lastly, histologists could use DL to detect presence of the PD-L1 biomarker from WSIs of cancerous tissues, and this finding could then be used by oncologists to recommend ICI therapy. All these findings could eventually be applied to detect/treat/predict outcomes of other types of cancer and even different conditions, and numerous lives could be potentially saved.
Searching for a relevant material for my capstone paper was a bit of a challenge. Although my article search included peer-reviewed journals between 2010 and 2020, all of the articles relevant to the subject, with the exception of three, were published in the last years. Also, the majority of articles that were yielded through my databases search were irrelevant. This shows that AI application is relatively novel in the healthcare field and has great opportunities to make improvements in healthcare, especially in pulmonary oncology and personalized medicine. Another limitation to my search may be that some research data may not have been published if the results were negative or equivocal. In that case, important data would not be available to be shared with the healthcare and research groups, which can even lead to the endangering life of the patients (Joober, Schmitz, Annable, & Boksa, 2012).
Some of the areas in medicine where AI could be used in the future for detecting, diagnosing, monitoring patient’s response to treatment, and even surgically treating lung cancer patients include teleradiology, telepathology, and telesurgery. By using AI in teleradiology, virtual image transfer between healthcare facilities will be facilitated. This will allow radiologists to expand access to care in underserved or underdeveloped areas, and AI could help with reading images if no available radiologists are present to do so (Farahani, Riben, Evans, & Pantanowitz, 2016). This would allow for faster detection of cancer for some patients, and these patients should be able to get customized treatment plans faster. Additionally, teleradiology could be used for comparing patient images and monitoring patient response to treatment.
In telepathology, AI can be used in a similar manner as in teleradiology. AI in telepathology would enable providers in underserved or underdeveloped communities to more easily share WSIs among pathologists and other providers for education, research, consultations, and most importantly diagnostic purposes (Hitchcock, 2011). This in turn could help with faster lung cancer detection and diagnosis for some patients and speedier establishment of a treatment plan.
Moreover, AI in the future could be used in telesurgery for oncology patients in rural and underserved areas, where there is typically a shortage of surgeons (Choi, Oskouian, & Tubbs, 2018). Surgeons who opt for telesurgery use wireless networking and robotic technology to perform surgeries remotely on patients and are also able to collaborate with other surgeons at different surgical sites in real-time (Choi et al., 2018).
Using telesurgery on oncology patients will enable those patients, who most likely already have a huge financial burden and are not in the best shape to travel, to avoid travel-related costs and risky long-distance travels, as well as faster recovery since damages to healthy tissues will be minimalized (Choi et al., 2018). Because of all these benefits, telesurgery should be considered for lung cancer patients, and surgeons should find a way to use telesurgery in oncology more often. These are just some of the main areas in the healthcare where AI use could be more expanded in the future and which should benefit oncology patients. Oncology definitively has a new chapter ahead when it comes to AI applications.