Cancer treatment has come a long way in recent years and is now evolving more rapidly through the integration of artificial intelligence (AI) tools, such as machine learning (ML). Currently, health data exists in many forms, including electronic health records (EHR), diagnostic images, genomic and molecular data, pharmacological data, and patient-reported data. The creation of state-of-the-art cancer treatments can be enhanced by the ways clinicians leverage data to optimize care, and there’s no better way to achieve this than through the use of AI.
Completing the picture
There is a missing element to most EHR, AI, and ML models today – real-time patient-reported outcomes. This type of data refers to information regarding patients' experiences with their medical conditions, treatments, and healthcare providers, collected from outside a controlled clinical trial setting. Continuous patient data from the time periods between clinic visits is proving to be a critical component that is absent from most EHR, AI, and ML models available today. While oncology AI solutions have made impressive strides, the lack of real-time data represents a crucial gap in the current landscape. More than ever, clinicians need robust AI solutions that allow regular input of patient experiences to create stronger data sets and identify opportunities for timely interventions.
A study published in the Journal of Clinical Oncology explored the impact of enhanced symptom monitoring and health-related quality of life on clinical outcomes during routine cancer care using patient-reported outcome measures. Patients undergoing routine chemotherapy treatments were asked to report symptoms via tablet computers or to receive usual care consisting of symptom monitoring at the discretion of clinicians. Those with home computers received weekly emails to report between visits. The study concluded significant clinical benefits of symptom self-reporting during cancer care, including increased rates of discussion between patients and clinicians and subsequent measures taken to manage symptoms in response to patient reports leading to more well-controlled symptoms
Continuous patient engagement provides advanced ML models with the data required to establish increased predictive capabilities, for example, to determine how symptoms might develop for certain treatments. With the right data, AI can help build a comprehensive view of how patients respond to treatments in real-world settings, including factors that may affect treatment outcomes but are often excluded from clinical trials, such as patient demographics, lifestyle, and co-morbidities.
Data collection — timing matters
Not only do cancer patients frequently contend with life-threatening conditions, but many of the treatments available such as radiation and chemotherapy, are highly toxic, potentially leading to serious side effects or adverse events if not monitored closely. The key to minimizing the impact of treatment toxicities is to enable clinicians to pick up on their development as early as possible by monitoring metrics, such as laboratory results, clinical data, and patient symptoms occurring throughout the treatment cycle, allowing prompt intervention when necessary.
A published review of the outlook of oncology informatics examined the utility of gathering high-quality data through digital patient-reported outcomes. There was a high acceptance among the general population and healthcare providers, showing that ongoing patient self-reports can simultaneously provide benefits to cancer patients, providers, and developing AI algorithms.
Most data available to healthcare AI platforms are from past experiences, surveys, or records. Without continuously refreshed data, oncology providers are often unable to offer the level of proactive and responsive treatment options that cancer patients require. Many clinicians and healthcare companies have begun to enhance their care strategies by implementing the latest AI and ML technologies to gather information in real time when it is most beneficial to patients.
A study conducted at the Oulu University Hospital in Finland demonstrated the ability of an ML model to predict patient responses to oncology treatment with 75% accuracy. In this study, an ML model predicted the objective response rate of patients undergoing immune checkpoint inhibitor therapies using clinical and patient-reported data. These promising results highlight a need for a larger dataset to showcase the true power of ML-based approaches in treatment-response prediction.
Options for increased collection of real-time experiences currently exist in the form of AI and ML platforms in patient-facing apps, allowing patients to report their symptoms, side effects, and other relevant health information, such as medication usage, adverse events, and quality-of-life indicators. These smart solutions, often combined with web-based dashboards for healthcare providers, enable a more comprehensive understanding of the patient experience by providing highly detailed overviews of patient-reported data throughout the treatment cycle.
There is a critical need for physicians to have access to real-time patient data from the point of diagnosis through treatment and follow-up to optimize patient outcomes. Platforms today are paving the way for highly personalized and efficacious cancer care by providing care teams with the ongoing data they require and giving comprehensive views of each patient’s treatment plan and symptom development at a glance. These advanced platforms transform data input into automated alerts when necessary, sending messages to care teams when interventions are required
Following a cancer diagnosis, patients who use these software platforms or applications to access educational resources and up-to-date information on symptom management and early intervention benefit from tailored resources based on their own input. As time progresses, these intelligent applications become more precise in predicting the development of symptoms and can provide guidance on how to alleviate them from home. Once the active treatment phase is completed, these systems can capture and track long-term outcomes, detecting symptoms that may emerge later, which is vital in alerting healthcare providers to any potential signs of relapse.
Limitless possibilities for personalized cancer care
Insights derived from the latest AI platforms are beginning to shape and inform the development of new treatments and clinical trials, leading to more personalized and effective cancer care. Through leveraging AI and ML technology, emerging health management platforms are actively enabling healthcare providers to maximize their operations and offer personalized, next-level care to each patient.
The use of AI and ML to collect and analyze real-time patient experience data in oncology is revolutionizing how we approach cancer care, guiding treatment decisions and ultimately improving patient outcomes.
Anish Patankar is the senior vice president and general manager of oncology software at Elekta. Patankar is a health tech expert specializing in the cloud, Internet-of-Things, big data, and artificial intelligence. From building leading-edge digital information solutions to deliver revenue growth, he has experience spanning digital health tech, medical devices, education, and security software industries.