By Colin Ruggiero
The internet of things (IoT), with its promise to “optimize every aspect of care and transform the way it is managed across the continuum,” is a technology that could reinvent healthcare. There is a growing demand within healthcare for more technology integration, as the combination offers clear convenience, efficiency, and automation (Shashank, 2017).
More precisely, the healthcare-focused subset of the IoT is called the internet of medical things (IoMT) (Alliance of Advanced BioMedical Engineering, 2017). The IoMT will offer improved healthcare by allowing for more accurate measurement of information. Collecting, storing, and reviewing information provides an overarching plethora of data that can be accessed by machines and analyzed for patterns.
The IoMT does not supersede doctors’ diagnoses or treatment plans, but it can assist healthcare professionals in confronting stalls or difficulties. All patients can benefit from technology that will lead to more informed decisions; however, cancer patients may benefit the most due to the importance of early intervention and personalized treatment.
While the IoMT produces its own comprehensive data, it can be used for artificial intelligence (AI). The IoMT lays the groundwork that AI employs: “[E]xisting IoT platforms [provide] the interface to gather the data from various devices and can offer a relatively easy way to utilize the IoT data [in] AI/ML systems” (Kubara, 2019). ML, or machine learning, is a component of AI, specific to the way data is studied, that deals with predicting outcomes and recognizing sequences from existing input.
Because AI does not require supervised learning, it makes the IoMT sharper (Schmelzer, 2019). AI employs clustering approaches (“a collection of objects on the basis of similarity and dissimilarity between them”) and ML systems are able to identify patterns (Priy, n.d.). From this, the process calls attention to any irregularities, whether they are abnormal or consistent. AI-enabled IoT systems pick out “relevant insights that might not be visible” through a “haystack of data” (Schmelzer, 2019).
By merging AI into the IoMT, this data process yields healthcare that is higher quality, more proficient, and more meaningful. A large part of AI is ML, which is becoming a prominent feature in healthcare.
Deep ML covers a range of applications. For cancer patients, ML assists radiologists—many times outperforming them—and has proven to be a resource through “computer-aided detection and diagnosis supporting clinical decision making and survival analysis” (Lundervold & Lundervold, 2019).
Lately, ML technology has been implemented for the detection of several cancers. In July 2019, the Dana-Farber Cancer Institute “developed a deep learning tool that performed as well as human reviewers in extracting clinical information regarding changes in tumors from unstructured radiology reports for patients with lung cancer” (Kent, 2019).
Lung cancer is one of the most common types of cancer worldwide. AI technology can provide augmented care for lung cancer patients by observing tumor growth and abnormalities, and doing so with extreme efficiency. The results showed that “[h]uman reviewers were able to annotate imaging reports for about three patients per hour,” while the deep ML used by Dana-Farber was able to annotate the reports for the entire cohort (totaling nearly 30,000 reports) in roughly 10 minutes (Kent, 2019).
A cancer type that may be misdiagnosed as lung cancer, mesothelioma, has also been receiving AI attention. In the fall of 2019, an AI model for mesothelioma patients identified potential cases and predicted the best treatments for them (Hale, 2019). Through applying the data processing of AI and the IoMT, Owkin researchers took 3,000 whole-patient biopsy slide images to indicate “novel biological features of the tumor and link them to its overall prognosis.”
Advancements in mesothelioma diagnosis have been long overdue. For decades, this rare form of cancer has primarily burdened the senior community. Because mesothelioma symptoms can mimic other, less severe conditions, patients are usually not diagnosed until stage 3 or 4 and have poor prognoses and life expectancy (Molinari, 2020). Now, MesoNet, the AI mesothelioma model, is improving detection and treatment for these patients who typically have a short life expectancy.
AI’s influence in healthcare is reaching wide populations. For breast cancer patients, this means more precise screenings. A persistent complication in breast cancer diagnosis is “the interpretation of mammograms [which are frequently] affected by high rates of false positives and false negatives,” according to an article in the journal Nature (McKinney et al., 2020). This publication, which has studied an AI system’s effectiveness compared to human experts, sheds light on another promising development.
After lung cancer, breast cancer is the second most common type of cancer (Breast Cancer Research Foundation, n.d.). Researchers found that the use of AI to predict breast cancer advancement through mammogram interpretation resulted in “an absolute reduction of 5.7% and 1.2% (USA and UK) in false positives and 9.4% and 2.7% in false negatives” (McKinney et al., 2020). It is crucial to remember that AI’s role in healthcare is minor in comparison to human judgment. Yet, this study demonstrated that the AI system “reduced the workload of [a human] reader by 88%” (McKinney et al., 2020).
Early cancer prognosis is achievable, especially as technology grows and integrates into modern healthcare. Even with these small beginnings, AI technology has already established a standard that can be used to measure success, which includes healthcare quality and patient safety.
Colin Ruggiero dedicates his time to informing others about mesothelioma and preventive measures to avoid exposure to asbestos.
Alliance of Advanced BioMedical Engineering. (2017). Internet of medical things revolutionizing healthcare. https://aabme.asme.org/posts/internet-of-medical-things-revolutionizing-healthcare
Breast Cancer Research Foundation. (n.d.). Breast cancer statistics and resources. https://www.bcrf.org/breast-cancer-statistics-and-resources
Hale, C. (2019, October 7). Owkin’s mesothelioma AI discovers new biomarkers from lung biopsy images. FierceBiotech. https://www.fiercebiotech.com/medtech/owkin-s-mesothelioma-ai-discovers-new-biomarkers-from-lung-biopsy-images
Kent, J. (2019, July 30). Deep learning tool detects cancer in radiology reports. HealthITAnalytics. https://healthitanalytics.com/news/deep-learning-tool-detects-cancer-in-radiology-reports
Kubara, K. (2019, July 10). Artificial intelligence meets the internet of things. Towards Data Science.
Lundervold, A. S., & Lundervold, A. (2019). An overview of deep learning in medical imaging focusing on MRI. Zeitschrift für Medizinische Physik, 29(2), 102–127. https://doi.org/10.1016/j.zemedi.2018.11.002
McKinney, S. M., Sieniek, M., Godbole, V., Godwin, J., Antropova, N., Ashrafian, H., … Shetty, S. (2020). International evaluation of an AI system for breast cancer screening. Nature, 577, 89–94. https://doi.org/10.1038/s41586-019-1799-6
Molinari, L. (2020, January 28). Mesothelioma cancer. Mesothelioma.com. https://www.mesothelioma.com/mesothelioma
Priy, S. (n.d.). Clustering in machine learning. GeeksforGeeks. https://www.geeksforgeeks.org/clustering-in-machine-learning
Schmelzer, R. (2019, October 1). Making the internet of things (IoT) more intelligent with AI. Forbes. https://www.forbes.com/sites/cognitiveworld/2019/10/01/making-the-internet-of-things-iot-more-intelligent-with-ai/#3c0930defd9b
Shashank, A. (2017, November 3). 6 reasons why healthcare needs the internet of things (IoT). HIT Consultant. https://hitconsultant.net/2017/11/03/internet-things-digital-future-value-based-care/#.XjwnpBNKhQJ