Two comprehensive review papers thoroughly document and summarize the ML research that has been performed for cancer in the last three decades. The first review paper is by the University of Alberta in Canada and provides a thorough review of the studies undertaken prior to it writing in 2006 (Cruz and Wishart, 2006). The second paper is a more recent one and comes from the University of Ioannina in Greece (Kourou et al., 2015). Both papers have broken down the types of subproblems in the area of cancer ML quite well and match well with some of the subproblems that were discussed in Chapter 2, Healthcare Foundations. These problems include:

• Early detection/screening of cancer: Can machine learning models be trained to identify individuals with a high risk of developing cancer before symptoms appear?
• Diagnosis of cancer: How can machine learning aid oncologists/radiologists in making definitive diagnoses of cancer and classifying the cancer stage and grade?
• Cancer recurrence: In an individual who has been treated successfully with initial cancer, how likely is that cancer to recur?
• Life expectancy and survivability: In which patients will cancer likely cause mortality? Which patients are likely to be alive in 5 years? In 10 years?
• Tumor drug sensitivity: Which tumors are likely to respond to specific treatments for cancer (for example, chemotherapy, radiotherapy, biological agents, or surgery).