Building machines capable of carrying out activities that need human intelligence falls under the broad umbrella of artificial intelligence, a branch of computer science. It often refers to computational innovations that mimic human intelligence-assisted processes like cognition, engagement, deep learning, sensory perception, and adaptation. Some machines can do tasks that usually require human interpretation and judgment calls. AI and similar technologies are becoming more and more common in society and business and are starting to be used in healthcare. These technologies could transform many facets of patient care and operational procedures within providers, payers, and pharmaceutical companies. Researchers and healthcare professionals are paying attention to artificial intelligence (AI) in the healthcare industry. Few prior research, including those in business, accounting, management, decision sciences, and the health professions, have examined this subject from a multidisciplinary angle. The application of AI in healthcare is expected to increase significantly. Gene editing, medication research, personalized medicine, supportive healthcare services, and illness diagnosis are some of the current applications. Diagnosis, discovery, and treatment planning have all been revolutionized by artificial intelligence (AI). In addition to helping with cancer detection, it can also help with cancer therapy design, finding novel therapeutic targets by speeding up drug discovery, and enhancing cancer surveillance by looking at patient and cancer statistics. AI-guided cancer treatment may improve clinical management and screening with more positive health outcomes. This thorough study offers a systematic assessment of the literature on using artificial intelligence (AI) in cancer diagnosis and provides crucial insights into the uses of AI for cancer treatment.
There are currently several cancer treatment options. Since the 2010s, cancer treatment has become significantly more effective. Despite the profusion of modern instruments, precise diagnostic therapies that are scientifically effective for each sick person are still challenging. Therefore, a patient-specific, optimal treatment may be used if an accurate diagnosis could be made. Improved forecast accuracy could aid doctors in better planning patient therapies and reducing the pain and suffering brought on by the condition. After treatment, cancer can return fast, and it can be challenging to detect in the early stages. Furthermore, making exact predictions of clinical diagnoses is exceedingly tricky. Some cancers are difficult to see in the early stages due to their vague symptoms and the hard-to-distinguish telltale indications on mammograms and scans. Therefore, developing better predictive models using multivariate data and cutting-edge diagnostic technology for clinical cancer research is essential. According to a quick literature study, AI is more precise than traditional analytical methods like data analysis and multivariate analysis. This is particularly true when cutting-edge bioinformatics tools are used with AI because this can significantly improve the precision of prognostic, diagnostic, and predictive techniques. A more specific idea known as machine learning (ML) is becoming more popular. To predict a patient’s chances of survival, prediction models are developed using machine learning (ML), a subset of artificial intelligence. Machine Learning learns logical patterns from a significant amount of historical information.
How is artificial intelligence applied in cancer diagnosis and treatment?
This review is drafted using contextual investigation examination of different and heterogeneous sources mainly academic journals, periodicals, websites and websites. The review permits a comprehensive understanding of the application of artificial intelligence in cancer treatment and diagnosis.
Malignant tumors are incredibly aggressive and endanger the lives of those they affect because of the fast cell multiplication, invasion of adjacent organs, and formation of new growths known as metastases. Tumors are now categorized using a variety of methods. For instance, experts can use medical imaging to remove Tissue and then study it under a microscope to determine how different healthy Tissue and malignant tissues can be distinguished (Huang et al.)[1]. Important information can be gleaned from the primary tumor’s size and location as well as the presence of distant metastases. Tumor markers can also be utilized in laboratory diagnostic procedures to demonstrate the existence of a tumor as per (Searching for the Causes of Cancer)[2]. Artificial intelligence-based picture identification technology has advanced significantly in recent years. Numerous researchers and businesses are now developing various approaches to increase the efficiency, precision, and affordability of cancer screening. While a few of these are just beginning, others are further along, like Paige. AI produces software for prostate and breast cancer that has the CE label. AI can serve as a second set of eyes for pathologists, enabling them to discover cancerous growths on a specimen they might otherwise miss. Better accuracy and overall improved patient insights may result from this. In environments with limited resources, AI can be beneficial for cancer screening. Compared to radiologists, an AI program can learn from a more comprehensive library.
Interest in AI applications in these crucial fields has grown over the past few years, sometimes with performance on par with human specialists and benefits in scalability and time-saving (Chen et al., 2021)[3]. According to (Chen et al.) the underlying epigenetic and genetic heterogeneity has also been described using histopathology pictures and Deep Learning techniques. Additionally, the authors argue that Deep Learning techniques predicted whole-genome duplications, chromosomal arm losses, gains in focal additions and deletions, and gene changes for pan-cancer have all been foretold. In addition to predicting mutations in specific genes, Deep Learning models have been used to forecast mutational footprints, which are the most significant biomarkers for responses to checkpoint immunotherapy. Examples of these mutational footprints include microsatellite instability (MSI) status and tumor mutational burden (TMB) status.
(Boniolo et al.) insists that one of the best instances of how AI has been utilized to produce tailored medicinal ingredients is personalized cancer vaccinations. Cancer vaccines need to identify antigen peptides that are extremely specific to the patient’s tumor and MHC genotype to strengthen the patient’s immune system. According to (Boniolo et al.)[4] most individualized vaccine design processes now incorporate optimization techniques and machine learning to help with peptide identification and vaccine assembly. Such vaccine design frameworks are advantageous for population-level preventive vaccine development against contagious diseases because they allow for selecting a target antigen and set of MHC alleles, making them amenable to customized cancer immunotherapy. Similarly, AI-driven medicine discovery for both large and small molecules has recently experienced results, with some examples of tailored applications.
Cancer immunotherapies use the immune system of the patient to combat malignancies. Unique genetic changes that give rise to neoepitopes, a class of self-peptides linked to the primary histocompatibility complex (MHC) and used to distinguish between cancerous and healthy cells, may occur during the evolution of tumors. Since then, AI has emerged as a crucial component of many cancer vaccine design pipelines, from predicting neoepitopes from a patient’s distinctively altered peptide pool to selecting and assembling the neoepitopes into vaccines. AI-based methods for personalized vaccine design have increasingly developed such cancer-specific peptides as per (Boniolo et al., 2021).
The use of AI in medicine development begins with AI is utilized at every stage of the current pipeline for cancer vaccines. The final vaccination is chosen and put together using constraint optimization models, which are used to find cancer-specific antigenic peptides. In recent years, generative models that use co-evolutionary data from protein sequences have changed structural biology. These models are currently used to predict 3D structures, speed up molecular dynamics simulations, and create new proteins. Deep neural networks are now revolutionizing the creation of new tiny compounds. With generative models, it is simple to develop novel molecules with enhanced biological properties while exploring a sizable percentage of the chemical search space (Boniolo et al., 2021).
AI is used to predict the effects of anticancer medications or to facilitate the discovery of anticancer treatments. Different cancers and drugs might react differently, and data from extensive screening procedures frequently shows a connection between the genetic variety of cancer cells and therapeutic efficacy. Researchers produced fake data using monitoring data and ML. The methodology is used to forecast the effectiveness of anticancer drugs based on the location of the present mutation in a malignant cell’s genome (Alqahtani, 2022)[5]. Moreover, AI holds great promise for determining an anticancer drug’s susceptibility (Alqahtani, 2022). AI plays a crucial role in fighting against cancer drug resistance. AI can quickly understand how cancer cells develop immunity to cancer treatments by analyzing data on large drug-resistant tumors, which can help advance drug development and usage.
The potential impact of artificial intelligence (AI) methods on several aspects of cancer therapy is extensive. These include the creation and development of pharmaceuticals and the clinical validation and eventual administration of these drugs at the point of care, among other things(Ho, 2020)[6]. At the moment, these procedures are costly and time-consuming. Furthermore, different patients experience different effects from their therapy. There are numerous strategies to deal with these issues due to the convergence of AI and cancer therapy. Machine learning and neural networks are the only AI platforms that can speed up drug discovery, use biomarkers to precisely match patients to clinical trials, and truly customize cancer treatment using only the patient’s data (Ho, 2020). Some medicines stimulate the body’s defense mechanisms to fight cancer cells, as with immunotherapy. The cancer cells themselves, however, target the three main pillars of modern cancer therapy: surgical removal of the tumor, chemotherapy, and radiotherapy. Treating cancer using radiotherapy involves using ionizing or particle radiation to damage the genetic material of cells to prevent them from dividing further (Iqbal et al., 2021)[7]. This has proven to be one of the most effective methods for shrinking or eliminating tumors since the 20th century. Although the radiation also damages healthy cells, these cells are unlike their diseased counterparts in that they are better able to repair themselves – depending on the severity of the damage. Whereas the cancer cells die off, the plan is for the healthy tissue to regenerate. For this reason, historically, radiation dose has been administered over several sessions, known as fractions—to give the healthy tissues time to repair themselves between treatment sessions. However, recent advances in precision that enable targeted treatments that better protect the surrounding healthy tissues have enabled more precise radiotherapy treatments, such as intensity-modulated and image-guided radiation therapy.
Lack of organized cancer-related health data and a lack of uniformity in the collection and storage of unstructured data inside an EHR or integrated database of a given healthcare system present significant challenges for the database creation of AI models (Alkhaldi, 2021)[8]. Because it restricts interoperability and the mass exchange of health information and data, the absence of uniformity across healthcare systems and governments worldwide is even more significant. To tackle this, the Minimum Common Oncology Data Elements program created standardized terminology and descriptions for the frequently used patient- and tumor-related features, disease status categories, and treatment interventions. Its use in ordinary clinical practice is still being investigated, and its application necessitates critical information technologies and systems resources.
Additionally, the “black box” aspect of the mechanism, which is frequently discussed, makes it difficult for medical professionals to accept AI technologies. This is especially true for systems based on deep learning and neural networks, which depend on complex hidden neurons of data interaction. Although simple ML algorithms, such as linear regression, operate perfectly transparently, many contemporary approaches use techniques that involve creating many overlaying decision trees with complex reinforcement schemes that cannot be usefully depicted graphically. DL, which is based on hidden layers of data interaction inspired by the interconnection of the neurons and synapses of the brain, further complicates interpretability (Shreve et al.)[9].
Low levels of expertise cause a challenge in understanding an AI-powered study hence becoming another roadblock to the adoption of AI in healthcare (Victor Mugabe)[10]. Instead of focusing on how the science is technically carried out, principal investigators should discover what topics AI is particularly well-suited to address. AI tries to model a complicated system and provide precise predictions, in contrast to traditional statistical methods used to evaluate correlations between variables and provide directed hypothesis testing. For widespread acceptance, there must be open communication between healthcare experts, the public, and developers (Albert, 2021)[11].
The most important factors influencing treatment choices and patient outcomes are cancer screening for timely detection, precise cancer diagnosis, classification, and grading. AI could be used for cancer screening in a variety of ways, including pre-screening to weed out patients with low cancer risk, replacing radiologists as readers entirely, replacing one radiologist in a configuration with two or more readers, assisting radiologists in making diagnoses, and adding an extra level of diagnostic testing upon basic radiology assessment. Each of these calls for a slightly different strategy for developing and testing AI technologies. More research and development are required before AI-based diagnostics are used widely in cancer testing and other medical fields.
For many years, surgery, chemotherapy, and radiotherapy will still be the conventional cancer treatments, but the scientific community is becoming increasingly interested in developing the current clinical cancer treatment approaches. According to research, after systematically analyzing data from large pharmaceutical and clinical datasets, AI is recognized as one of the top futuristic therapies for accurate cancer diagnosis, prognosis, and treatment. The application of algorithm-based AI support for radiology image processing, data mining, and electronic health records, to give a more precise answer for cancer therapy is anticipated to alter clinical practices and digital healthcare in the future. The impact of ML on healthcare procedures is significant. It may impact diagnosis and therapy, raising important ethical questions. Applications for machine learning in healthcare span from completely autonomous AI for cancer diagnosis to nonlinear mortality estimates to help with resource allocation.
The prediction of cancer can benefit from AI and machine learning. Artificial intelligence can spot malignancies that have already spread and people at a high risk of getting it before it does. Medical professionals monitor these patients closely and act quickly when necessary. Additionally, ML in cancer detection can assist people in receiving preliminary feedback on their anomalies through self-diagnosis without needing a doctor’s appointment. Such tools do not always result in a conclusive diagnosis. The approval of pathologists is required. The same medicine may respond differently to various forms of cancer. AI can forecast how different drugs would affect malignant cells. This information aids in the creation of new anticancer medications and the timing of their use. Creating individualized treatments and tumor characterization can benefit from improving genome sequencing with artificial intelligence.
The detection and treatment of cancer can significantly benefit from artificial intelligence. To begin with, artificial intelligence makes room for tailored cancer therapy procedures. Big data and AI enable medical professionals to examine various data about the patient and the cancer cells to develop individualized treatments. The side effects of this kind of therapy will be less severe. Less harm will be done to healthy cells, but it will significantly affect cancer cells. Second, the use of artificial intelligence in the detection and treatment of cancer enhances diagnostic precision by lowering false-positive and false-negative results. Proof comes from studies on breast cancer detection. One in ten female patients with mammograms examined by doctors has false-positive results, forcing them to undergo stressful procedures and unnecessary invasive testing. The research team at Google created software that uses AI to reduce false positive and false negative mammography readings by 6% and 9%, respectively. Another group of researchers developed an AI algorithm to identify breast cancer. This algorithm assisted radiologists in lowering false-positive rates by 37.3% during an examination.
Additionally, the development of artificial intelligence has made it possible to identify cancers without invasive techniques. Sometimes the tumor’s benign nature is discovered only after the removal surgery, which would have allowed the procedure to be avoided entirely. Such occurrences can be significantly decreased with AI’s assistance in the cancer detection process. Image-guided needle biopsies can train machine-learning algorithms to recognize malignant tumors. Without the need for intrusive procedures that would otherwise be necessary, they can identify and characterize the isocitrate dehydrogenase (IDH) mutations from MRI imaging of gliomas. In the end, artificial intelligence is crucial in preventing cancer overtreatment by assisting radiologists in determining which tumors and anomalies are malignant and require actual treatment. Artificial intelligence (AI) systems can identify precancerous lesions in photos and separate them from other abnormalities, preventing over-treating patients for unimportant conditions.
Artificial intelligence has the potential to save both patients’ lives and physicians’ time. Early cancer diagnosis could be revolutionized by using AI in healthcare data, and automation could support concerns about capacity. AI may make it possible to analyze complex data from various sources, including radiomic, genomic, metabolomic, and clinical text data. It is envisaged that ongoing research to assist the application of AI to cancer genomes would enable multicancer early detection and tumor site origin determination. This may improve cancer survivors’ surveillance plans and change cancer screening, especially for less common and rare tumors. The difficulties in designing, implementing, and maintaining AI models that have been raised are significant but not insurmountable. Initial AI tools now becoming accessible within the EHR are helping cancer practitioners, and there is a clamor for their potential future applications.
Albert, Helen. “AI for Cancer Detection: Ready for Prime Time or Caution Advised?” Inside Precision Medicine, 24 Sept. 2021, www.insideprecisionmedicine.com/artificial-intelligence/ai-for-cancer-detection-ready-for-prime-time-or-caution-advised/?gclid=Cj0KCQjw9ZGYBhCEARIsAEUXITXO-btdwuec8pxaZJdr9lU8rdK5jmrQDdf5NQeXNG4–VhTq-HlXfIaAnWlEALw_wcB.
Alkhaldi, Nadejda. “AI in Cancer Detection and Treatment: Applications, Benefits, and Challenges.” ITRex, 22 Oct. 2021, itrexgroup.com/blog/ai-in-cancer-detection-treatment-applications-benefits-challenges/#header.
Alqahtani, Amal. “Application of Artificial Intelligence in Discovery and Development of Anticancer and Antidiabetic Therapeutic Agents.” Evidence-Based Complementary and Alternative Medicine, edited by Arpita Roy, vol. 2022, Apr. 2022, pp. 1–16, https://doi.org/10.1155/2022/6201067.
Boniolo, Fabio, et al. “Artificial Intelligence in Early Drug Discovery Enabling Precision Medicine.” Expert Opinion on Drug Discovery, June 2021, pp. 1–17, https://doi.org/10.1080/17460441.2021.1918096.
Chen, Zi‐Hang, et al. “Artificial Intelligence for Assisting Cancer Diagnosis and Treatment in the Era of Precision Medicine.” Cancer Communications, vol. 41, no. 11, Oct. 2021, pp. 1100–15, https://doi.org/10.1002/cac2.12215.
Ho, Dean. “Artificial Intelligence in Cancer Therapy.” Science, vol. 367, no. 6481, Feb. 2020, pp. 982–83, https://doi.org/10.1126/science.aaz3023.
Huang, Shigao, et al. “Artificial Intelligence in Cancer Diagnosis and Prognosis: Opportunities and Challenges.” Cancer Letters, vol. 471, Dec. 2019, https://doi.org/10.1016/j.canlet.2019.12.007.
Iqbal, Muhammad Javed, et al. “Clinical Applications of Artificial Intelligence and Machine Learning in Cancer Diagnosis: Looking into the Future.” Cancer Cell International, vol. 21, no. 1, May 2021, https://doi.org/10.1186/s12935-021-01981-1.
“Searching for the Causes of Cancer.” Www.siemens-Healthineers.com, 30 July 2021, www.siemens-healthineers.com/perspectives/causes-of-cancer.
Shreve, Jacob T., et al. “Artificial Intelligence in Oncology: Current Capabilities, Future Opportunities, and Ethical Considerations.” American Society of Clinical Oncology Educational Book, no. 42, July 2022, pp. 842–51, https://doi.org/10.1200/edbk_350652.
Victor Mugabe, Koki. “Barriers and Facilitators to the Adoption of Artificial Intelligence in Radiation Oncology: A New Zealand Study.” Technical Innovations & Patient Support in Radiation Oncology, vol. 18, June 2021, pp. 16–21, https://doi.org/10.1016/j.tipsro.2021.03.004.
[1] Artificial Intelligence in Cancer Diagnosis and Prognosis: Opportunities and Challenges
[2] Searching the Causes of Cancer
[3] Artificial Intelligence for Assisting Cancer Diagnosis and Treatment in the Era of Precision Medicine
[4] Artificial Intelligence in Early Drug Discovery Enabling Precision Medicine
[5] Application of Artificial Intelligence in the Discovery and Development of Anticancer and Antidiabetic Therapeutic Agents
[6] Artificial Intelligence in Cancer Therapy
[7] Clinical Applications of Artificial Intelligence and Machine Learning in Cancer Diagnosis: Looking into the Future
[8] AI in Cancer Detection and Treatment: Applications, Benefits, and Challenges
[9] Artificial Intelligence in Oncology: Current Capabilities, Future Opportunities, and Ethical Considerations
[10] Barriers and Facilitators to the Adoption of Artificial Intelligence in Radiation Oncology: A New Zealand Study
[11] AI for Cancer Detection: Ready for Prime Time or Caution Advised