Cancer is one of the top causes of death worldwide. According to the American Cancer Society, 609,360 cancer deaths and 1.9 million new cancer cases were expected to occur in the US in 2022 alone.
Due to terminal disease prevalence and increased patient risk, scientific, technological, and AI advancements are likely to progress faster in oncology than in other chronic diseases.
AI in oncology will address critical oncology challenges
Today, Artificial Intelligence (AI) has tremendous potential in oncology, promising to:
- Assist in genomic analysis, clinical trial design, drug development, etc.
- Analyze various oncology-related data types, including proteomic, metabolomic, transcriptomic, and other relevant data.
- Identify early cancer diagnosis through accurate cancer detection.
- Easily develop individualized therapies.
3+1 ways ΑΙ is reshaping Oncology
Cancer imaging, clinical outcomes, translational oncology, and clinical decision-making are all areas transformed by AI in oncology.
- Cancer imaging
Deep Learning (DL) architecture and the Convolutional Neural Network (CNN) have transformed image analysis.
In skin imaging, clinical skin images captured with cutting-edge imaging technology demonstrate the potential of DL to detect skin cancer. For instance, Esteva et al. used DL to predict the histology and classification of skin lesions.
In radiographic imaging, AI helps in identifying the radiographic anatomic characteristics of malignancies. Clinicians historically met difficulty in diagnosing extranodal extension (ENE) radiographically. A CNN-based model demonstrated >85% accuracy in detecting ENE of tumors in head and neck cancer lymph nodes on diagnostic, contrast-enhanced CT images. This model holds potential as a clinical decision-making tool because ENE identification is crucial for the prognosis and treatment of patients with head and neck cancer.
Moreover, DL algorithms at digital pathology services contribute to characterizing the correlation of underlying genotype-phenotype in a tumor specimen.
- Clinical Outcomes
AI helps to improve understanding of clinical outcomes and disease development since it uses natural language processing techniques. DL algorithms can predict critical characteristics of malignancies, like survival rates, susceptibility, and the likelihood of recurrence.
Carrara et al. used an artificial neural network to predict late radiotherapy toxicity in patients with prostate cancer, and Lao et al. used brain MRI images and a DL approach and accurately predicted the survival rate.
- Translational Oncology
Translational oncology is emerging to demonstrate the efficacy of AI. DL neural networks have been employed to determine the protein structure and categorize cells into various mitotic stages that can further examine the lineage of progenitor cells useful to know the progression of the disease.
Clinicians may easily identify new cancer treatments using AI and genetic data, repurpose existing treatments and develop new effective ones against cancer. Menden MP et al. combined genomic and chemical characteristics to utilize an Artificial Neural Network (ANN) to predict the sensitivity of cancer cells to treatments.
Employing DLs, Feng et al. showed drug-target interaction strength prediction. Similarly, Eulenberg et al. utilized the DL method for predicting cell cycle reconstruction and disease progression.
- Clinical Decision making
Applications match patients with suitable clinical trials around the country by connecting their data to clinical trial databases. Also, machine learning (ML) is used to choose the best experimental medications for a specific patient. There has also been interest in using AI with patient data and national treatment recommendations to direct cancer management. IBM’s Watson for Oncology (WFO) is the most notable example. For patients with breast cancer, WFO has shown strong agreement with tumor boards’ recommendations but has fallen short in other areas of oncologic decision-making.
From cancer genetic research to patient relief, AI healthcare applications for cancer change the game.
Below we discuss examples showing how AI has transformed oncology for healthcare professionals and patients to tackle cancer.
Firecloud ensures large data access and analysis and facilitates research community collaboration.
Based on cloud computing infrastructure and supported by the elastic compute capacity of the Google cloud, FireCloud is one of the three NCI cloud Pilots.
It can be used to create, analyze, store, and share real-time datasets to identify diagnoses, regimens, and treatment patterns.
Powered by Terra, which uses an analytical platform to collaborate with cancer genomics research, Firecloud manages and computes large cancer research datasets ensuring ease of availability to encourage collaboration and enable reproducibility.
Firecloud’s collaboration with the Broad Institute and Watson Health to study cancer patients’ resistance and relapsing pattern to drugs shows how it can contribute to better cancer patient care.
Cancer Dost: The bot that answers cancer patients’ questions.
An AI-enabled chatbot developed to handle diagnosis, treatment, lifestyle management, and patient education is the most customized approach in digital oncology for patient care.
One of the most popular AI-enabled chatbots, available in English and Hindi, is Cancer Dost, developed by Hospido. Cancer Dost connects patients with physicians to get counseling over the phone or schedule in-person consultations. Meanwhile, patients get free access and, in a user-friendly layout, learn more about their disease, its severity and line of treatment, the organs affected, and their health condition.
As AI opens the door to more efficient cancer management and treatment, it is no surprise that discussions on the role of AI in oncology took place at ESMO2022.
ESMO highlights on AI oncology applications
Recently at ESMO Congress 2022, presentations and posters focused on how AI-driven tools will contribute to better cancer care.
For example, AI predicts immunotherapy efficacy in patients with non-small-cell lung cancer (NSCLC) and the prognosis of patients with high-grade serous ovarian carcinoma.
A Sanofi presentation highlighted two research collaborations the company formed with other companies to make the most of next-generation AI-based de novo design. Sanofi signed a research collaboration with Aqemia for a drug discovery project by generating its data with quantum and statistical physics-based calculations with the help of AI-based design of optimized molecules, and another one with Exscientia to integrate AI at the center of the Design-Make-Test-Analysis (DMTA) cycle.
Besides Sanofi, Achilles Therapeutics utilizes an AI platform validated with two types of real-world patient data. Identifying T cells reactive to predicted clonal neoantigens in patients’ samples validates the PELEUS platform. Over 120 patients have been prospectively analyzed to date across multiple cancer types. TRACERx study provides access to one of the largest longitudinal patient data sets ever created. It has collected 3,200 tumor regions from 795 NSCLC patients over five years.
Overcoming AI in Oncology limitations and heading to the future.
As they say, “nothing is perfect in God’s perfect plan,” and the same holds for technology. AI offers many benefits and has the potential to change the cancer industry. However, it comes with some limitations.
- Applying AI to disease processes with low prevalence can be challenging because DL neural networks need large data sets.
- Confidentiality issues come up, and concerns about disclosing protected patient health information.
- Although the DL model may accurately forecast the type and progression of cancer, it is difficult for us to understand the precise reasoning behind these predictions; this is the so-called “black box” problem.
What role will AI play in the future of oncology?
Despite certain obstacles, AI has demonstrated positive results in cancer imaging diagnostics, evaluation of treatment response, clinical outcomes prediction, and automated drug discovery process. We also see growing interest in AI applications in oncology at influential conferences (ESMO).
The notable amount of clinical data, including the entire genomes of patients generated over the past decade, proves AI’s power to transform oncology.
Addressing the challenges mentioned above, AI has the potential to transform oncology. Using diverse data, AI will drag cancer care into the 21st century and beyond.
Do you want to stay updated on the latest cancer treatment options and digital oncology?
Contact us and gain best-in-class market insights. Learn how we help you make informed decisions.
#Oncology #Digitalhealth #Cancer #digitaloncology #AIoncology #cancertreatment #digitalhealthinnovation #AIhealthcare #ESMO2022 #ESMO22
Sources:
Artificial Intelligence and Data Science – The future of oncology (softwebsolutions.com)
Technology can revolutionize cancer care. 6 experts explain how | World Economic Forum (weforum.org)
Artificial Intelligence (AI) in Oncology – Anesthesia Services (xenonflorida.com)
Artificial Intelligence in Oncology: Current Applications and Future Directions (cancernetwork.com)
[1606.05718] Deep Learning for Identifying Metastatic Breast Cancer (arxiv.org)
A robot will likely assist in your future surgery | Computerworld
Surgical Robotics Market Size to Hit US$ 21.3 Billion by (globenewswire.com)
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8474080/