Revolutionizing Outpatient Oncology Care: The Seamless Integration of Artificial Intelligence

Précis

Artificial intelligence holds immense potential to address the rising demand for oncology services and improve patient outcomes by facilitating more effective, efficient, personalized cancer care.

Abstract

With a shortage of oncologists predicted and an increase in demand for oncology services amid a growing aging population, the need for innovation is crucial. Artificial intelligence (AI) presents a promising solution, offering enhanced efficiency in various aspects of cancer care, from analyzing imaging to personalized treatment planning. The American Oncology Network (AON) has taken the initiative in integrating AI, utilizing platforms such as Flatiron Assist and VieCure to aid clinicians in decision-making processes, treatment planning, and patient management. Further, AON addresses challenges in data management with its subsidiary Meaningful Insights Biotech Analytics (MiBA). By transforming unstructured oncology data into actionable insights using AI, MiBA supports various clinical functions and helps optimize therapies. Despite the benefits, challenges such as data privacy, algorithm development, and ethical considerations remain. Ensuring the quality and integrity of data used for AI algorithms is crucial, as is addressing concerns regarding accountability and algorithm bias. Looking ahead, AI holds immense potential to revolutionize oncology care by improving diagnostic accuracy, treatment efficiency, and patient outcomes. Collaboration, transparency, and ongoing monitoring are essential for the successful integration of AI into clinical practice, with the ultimate goal of delivering more effective, efficient, and personalized cancer care.

Introduction

As research and detection methods have improved over time, more patients are receiving cancer diagnoses. In addition, patients with cancer are generally living longer than in previous decades, resulting in increasing demand for oncology services. As individuals age, the likelihood of encountering health challenges, including the possibility of a cancer diagnosis, increases. Further, it is estimated that the number of individuals in the US 65 years or older will double between 2000 and 2030, contributing to an increased demand for oncologists.¹

A study commissioned by the American Society of Clinical Oncology’s (ASCO) board of directors in 2007 revealed that demand for oncology services was expected to rise by 48% between 2005 and 2020; however, the supply of oncologists was expected to fall behind this, growing only 14% in the same time frame.¹ This equates to a shortage of approximately 2550 to 4080 oncologists, and this does not include the additional expected decrease in other oncology-trained professionals such as nurses and pharmacists. Results of an updated analysis of this ASCO study in 2014 showed the predicted demand for oncologists has remained consistent, hovering between 40% and 48% growth.²

As the need for oncology services continues to rise, the need for innovation and improved practice efficiency becomes increasingly imperative. Integration of artificial intelligence (AI) in oncology practices is one rapidly evolving strategy to improve efficiency in patient care.^3,4

With more than 240 providers in more than 80 community oncology clinics in 21 states, the American Oncology Network (AON) is in a unique position to leverage AI to help oncologists practice more efficiently and meet the growing demand for oncology services across the country. Flatiron Assist aids decision-making on the widely used OncoEMR platform, and VieCure is the sole patented platform in the oncology world with integrated AI. Additionally, Meaningful Insights Biotech Analytics (MiBA), an AON subsidiary, is revolutionizing data management by harnessing AI capabilities.

Reasons for AI Integration

Cancer care is becoming increasingly complex. Innovation and AI integration in outpatient oncology care can revolutionize the way cancer is diagnosed and treated. This integration can introduce advanced algorithms and machine-learning models that can analyze medical images, patient histories, and genomic data with exceptional precision. AI systems can detect patterns and correlations that might be overlooked by humans, potentially identifying cancers earlier and predicting disease progression more accurately. Furthermore, AI-powered tools are enabling the development of personalized treatment plans, tailored to the unique genetic and molecular profile of each patient’s cancer. Additionally, an enhanced ability to deliver precision medicine to patients improves outcomes and can reduce unnecessary treatments and associated adverse effects.^5-7

By enhancing diagnostic accuracy and treatment efficiency, AI has the potential to significantly increase survival rates and improve patients’ quality of life. Additionally, AI can help streamline administrative and clinical workflows, reducing the burden on health care professionals and allowing them to dedicate more time to direct patient care. AI can also automate routine tasks such as appointment scheduling, medication reminders, and patient follow-ups, thereby increasing clinician and staff productivity.⁸

Another significant complexity in oncology is the management and utilization of vast amounts of data generated in oncology care. AI can help harness these data for better decision-making, the advancement of research, and better-informed policy decisions.^9-11

AI Integration in the Electronic Medical Record

AON is steadfast in its commitment to advancing cancer care and treatment through cutting-edge technology and innovation. Among its initiatives, AON integrates AI into various facets of its operations.

Flatiron Assist

Flatiron Assist is revolutionizing decision-making within the OncoEMR platform, the most used platform at AON. Flatiron Assist leverages sophisticated machine-learning algorithms to analyze vast amounts of patient data, including electronic health records and genomic information. This analysis support assists oncologists in making informed clinical decisions. Oncologists can quickly select evidence-based treatment options at the point of care while clinical data for authorization and reimbursement are collected—all without disruptions to their day-to-day workflows.⁶

Looking ahead, AI holds tremendous potential to streamline processes such as selecting biosimilars based on organizational and insurance preferences. Currently at AON, this is an intense labor process that involves financial counselors and regional clinical pharmacists.

VieCure

VieCure’s integrated AI guides providers with diagnoses and genomic testing. AI analyzes the data and identifies missing tests, prompts necessary imaging studies, and guides therapy decisions. VieCure’s AI assistance extends further, processing patient-reported toxicities and laboratory results while providing actionable recommendations for treatment adjustments or alternative therapies. VieCure’s AI also aids in suggesting patients who should be included in clinical trials.⁵

MiBA

One significant challenge facing the health care sector is the large volume of oncology data housed within electronic medical record (EMR) systems, often rendered unusable in its current form. To address this, AON established MiBA, a forward-thinking data company. Leveraging AI, MiBA transforms these data into easily interpretable information for health care providers.⁷

MiBA’s operations involve extracting both structured and unstructured data from EMR systems within AON. Through machine learning algorithms, unstructured data (including PDF scans and external documents) are curated, rendering the data structured and actionable. These curated data are now easy to understand and serve diverse purposes across the system. As an example, these data populate clinical pharmacist dashboards, a tool used by pharmacists to optimize therapies based on specific clinical characteristics for guideline adherence or cost reduction. Furthermore, MiBA is poised to support clinical trial matching based on patient data, treatment history, and available clinical trials. Although this feature is under development, it represents an exciting frontier in cancer care innovation.

Challenges and Considerations

Although the potential for AI in oncology practice has its benefits, several challenges have limited its integration into practice. Primary concerns for clinicians include data privacy and security. Health care records are often a target for hackers due to their importance and vulnerability.⁹ Providing access to a third-party AI system raises concerns about potential Health Insurance Portability and Accountability Act violations.

In addition, setting up the algorithms for AI to use is a tedious process. These algorithms need to be reviewed and validated by skilled teams composed of medical experts in the oncology field. Although platforms such as MiBA can help with these types of tasks, not all organizations have the bandwidth to support a data integration company similar to MiBA. For AI to work at its full capacity, human clinical experience must be applied to label data correctly so that algorithms can utilize the information appropriately to make treatment decisions.

Several ethical concerns arise when dealing with AI, but the main concern is accountability. If AI leads to a wrong decision based on its algorithm, who is held responsible? To help with this, the FDA has investigated and started to develop standard guidelines for the moral use of AI in health care.⁹

Another ethical concern is algorithm bias. Some experts in the oncology field believe that AI will persuade providers to use a specific treatment more frequently than others. If used for the wrong purposes (ie, to gain market share for a specific drug), this can have a negative impact and definite ethical implications. AI platforms developed for health care professionals should help to identify appropriate therapies for patients based on FDA-approved indications, clinical guidelines, cost-effectiveness, and other clinical data. However, AI should not promote a treatment option less favored by available evidence and clinical guidelines.

One of the largest barriers to integration into clinical practice is the training required for providers to use clinical decision software to maximize its benefits. Due to the demands of daily operations, providers may be less inclined to undergo the appropriate training. Others may feel threatened by this technology for fear that if they adopt AI into their practices, their job will become obsolete. For AI to perform at its highest potential, appropriate training and adoption of this technology must occur at the clinician level.

Conclusion

As the prevalence of cancer among an aging population increases, the complexity of treatment selection is likewise increasing. AI stands to improve consistency and efficiency in the review and application of imaging, pathology, molecular studies, and other clinical information relevant to the optimization of cancer therapy.

The increasing complexity of cancer treatment selection is largely due to a rise in molecular testing and the expansion of targeted therapy vs traditional chemotherapy. Thanks to AI’s capability to enhance treatment selection, a greater proportion of patients with malignancies susceptible to targeted therapy will be identified and treated with the most appropriate therapy.

The importance of collaboration, transparency, and ongoing monitoring of AI tools cannot be underscored enough. Providers and other clinical staff, including pharmacists and nurses, must work closely with informatics teams to ensure that AI-based tools are both accurate and useful. Data that AI pulls from must be accurate, relevant, and timely to improve patient care and outcomes effectively. For continued improvement of AI capabilities, providers must continue to provide insight and feedback regarding how AI-based tools improve or fail to support patient care and process efficiencies. These data must be collected and published to promote the implementation of effective AI-based tools. Continuous updates and refinement are necessary as new data, testing, and targeted therapies emerge.

The prospects of AI-driven advancements in oncology are very exciting and will likely continue to improve patient care and outcomes. As data important for diagnosis and therapeutic selection become rapidly more complex, the ability to make these data available to providers in a timely and organized fashion becomes increasingly important, and AI-driven advancements will help us accomplish this in oncology.

REFERENCES

1. Erikson C, Salsberg E, Forte G, Bruinooge S, Goldstein M. Future supply and demand for oncologists: challenges to assuring access to oncology services. J Oncol Pract. 2007;3(2):79-86. doi:10.1200/JOP.0723601

2. Yang W, Williams JH, Hogan PF, et al. Projected supply of and demand for oncologists and radiation oncologists through 2025: an aging, better-insured population will result in shortage. J Oncol Pract. 2014;10(1):39-45. doi:10.1200/JOP.2013.001319

3. How will AI impact the future of clinical decision support? Flatiron. May 2023. Accessed April 19, 2024. https://flatiron.com/resources/how-will-aiimpact-the-future-of-clinical-decision-support

4. Debunking the myths of artificial intelligence in cancer care. Flatiron. December 2023. Accessed April 19, 2024. https://flatiron.com/resources/debunking-the-myths-of-artificial-intelligence-in-cancer-care

5. Halo Intelligence. VieCure. Accessed April 19, 2024. https://www.viecure.com/the-platform

6. Surface preferred regimens at the point of care. Flatiron. Accessed September 17, 2024. https://flatiron.com/oncology/clinical-decisionsupport

7. MiBA: a new era in data-driven oncology insights for a healthier tomorrow. MiBa Analytics. April 1, 2024. Accessed September 17, 2024. https://www.mibanalytics.com/press-release-miba-pioneering-data-drivenoncology-insights-for-a-healthier-tomorrow

8. Ashbury FD, Thompson K. Accelerating personalized medicine adoption in oncology: challenges and opportunities. In: Çalıyurt KT, ed. Accounting, Finance, Sustainability, Governance & Fraud: Theory and Application. Springer; 2023:41-49. doi:10.1007/978-981-99-5964-8_4

9. Khan B, Fatima H, Qureshi A, et al. Drawbacks of artificial intelligence and their potential solutions in the healthcare sector. Biomed Mater Devices. 2023;8:1-8. doi:10.1007/s44174-023-00063-2

10. Big majority of doctors see upsides to using health care AI. American Medical Association. January 12, 2024. Accessed September 17, 2024. https://www.ama-assn.org/practice-management/digital/big-majoritydoctors-see-upsides-using-health-care-ai

11. Ahuja AS. The impact of artificial intelligence in medicine on the future role of the physician. PeerJ. 2019;7:e7702. doi:10.7717/peerj.7702

The authors have nothing to disclose.