Evaluating Patient and Clinician Acceptance of AI Tools in Oncology: Lessons from Recent Research Findings

Cancer diagnosis and treatment require careful evaluation of many factors. These include tumor type, genetic markers, possible side effects, and patient preferences. Machine learning (ML) and AI systems are being made to help doctors manage these complicated tasks. Two recent research projects stand out for what they show about how AI tools work in oncology and how patients and doctors respond to them.

At Cornell University, researchers created an AI-guided decision support system called i-SDM (interactive Shared Decision-Making). It helps older adult cancer patients and their doctors talk better about treatment options, risks, and benefits. At the same time, Harvard Medical School scientists built CHIEF (Clinical Histopathology Imaging Evaluation Foundation). This AI platform looks at cancer tissue images to diagnose cancer, predict patient survival, and suggest treatment based on molecular profiles. Both tools show different uses of AI in oncology. i-SDM focuses on supporting decisions with patient involvement. CHIEF looks at improving diagnosis and understanding genetics.

Patient and Clinician Acceptance of AI Tools in Oncology

How well patients and doctors accept AI tools is very important for using AI in oncology. If people resist or doubt the tools, even the best technology might not work well. Recent studies from the US give detailed information about this issue.

AI and Older Adult Cancer Patients: i-SDM Experience

The Cornell study had good results for improving talks between doctors and older adults with cancer. The i-SDM system uses machine learning and language models to show treatment information made for each patient. This includes survival rates and side effects. It helps patients be involved and helps doctors understand concerns beyond medical facts, like travel costs and easy access to treatment. For example, older patients often must think about how far they travel for cancer treatment. This can affect their choices.

Still, some patients and doctors were unsure about using AI. Two patients and one doctor in the study did not want to use the tool. They worried the AI might not know enough and could give wrong information. This shows a big challenge: not everyone feels safe relying on AI for serious choices like cancer treatment. Older adults who don’t use technology much may feel this way more.

The research listed ten things that affect treatment choices. These include survival rates, risks, treatment options, costs, and travel. With AI help, doctors could talk about treatments better and with more facts. But shared decision-making takes more time and can be hard for some patients to understand during regular clinic visits. AI tools like i-SDM try to fix these problems, but people must trust them first.

Co-author Bob Riter said shared decision-making “accurately reflects patient wishes,” especially in cases like early-stage prostate cancer where many treatment options exist. Still, the concerns about AI show that the tools need to get easier to use and that patients need better education, especially older ones.

AI Diagnostic Tools and Broad Clinical Use: The CHIEF Platform

At Harvard Medical School, the CHIEF model is for a different kind of AI use — automatic diagnosis and predicting outcomes. This tool studies images from tumor biopsies to find cancer types, guess survival chances, and find genetic mutations that affect treatment. CHIEF was almost 94% accurate in finding cancer in 19 types. It could predict survival better than other models by 8-10%.

Because it is accurate and works for many cancer kinds, CHIEF might change cancer care workflows. It can work with images from different cancer types and hospitals, showing it can fit into real clinical work. Researchers pointed out that CHIEF found new tumor features, like immune cells linked to longer survival. Pathologists had not seen these before. This shows that AI can add to clinical knowledge.

But not all doctors accept new technology fast. Even if AI helps make better diagnoses, doctors need to trust and understand its advice before using it in treatments. Lead researcher Kun-Hsing Yu said they built a “nimble, versatile ChatGPT-like AI platform” meant to be used widely in clinics.

Challenges in Clinical Adoption of AI in Oncology

• Time Limits in Clinics: Oncologists often have little time with each patient. This makes it hard to explain AI results and make decisions together.
• Technology Skills of Patients: Older adults may not know how to use digital tools well. This makes AI advice seem confusing or cold.
• Worry About Errors and Gaps: Some patients and doctors fear AI might simplify or misread complex medical data, causing wrong ideas.
• Health Fairness Issues: AI tools must work well for many patient groups. Factors like income, race, ethnicity, and other health problems affect treatment and must be considered.

The Cornell team is working to add a health fairness part in i-SDM. This aims to help vulnerable groups who often have more health problems and less access to technology.

Getting approval from regulators and proving clinical value are also important. Some machine learning models have gotten official approval for certain cancer uses, which helps doctors trust them more. But wide use needs ongoing teaching and proof that AI helps in real care.

AI Integration and Workflow Automation in Oncology Practices

For cancer clinics and medical managers, adding AI tools means fitting them into daily work smoothly. AI should not make work harder but should help make tasks easier, improve work output, and support patient care.

Workflow Enhancement through AI Automation

• Automatic Routine Tasks: AI can handle front-office jobs like setting appointments, reminding patients, and taking first patient information. This lowers paperwork and lets staff spend more time with patients.
• Decision Support Systems: Tools like i-SDM give AI help for decisions by showing treatment choices clearly. This saves doctor time and makes talks with patients better. It supports shared decision-making when possible.
• Data Integration: AI tools like CHIEF study lots of image and molecular data fast. They give doctors important information quicker than reading it by hand.
• Helping Communication: AI answering services and phone systems can help cancer offices handle many patient calls about questions, test results, and appointments without stressing staff.
• Predictive Analysis: AI tools help sort patients by risk and predict results. This helps care teams decide what to do first and plan follow-up.

Considerations for Practice Implementation

• Staff Training: Clinic workers need good training to understand AI results and explain them clearly to patients.
• Patient Education: Patients should learn how AI helps so they feel better about using it.
• Data Security and Privacy: Following laws like HIPAA and keeping patient data safe when using AI is critical.
• Customization: AI tools should fit the needs of each oncology clinic, considering patient types and clinical goals.

IT managers in cancer centers should work closely with AI suppliers to make sure systems work well with electronic health records (EHRs) and that data flows smoothly.

Impact on Patient Engagement and Compliance

When AI tools give clear and useful information, patients tend to take part more in their care. This is very important in cancer treatment, which often lasts a long time and requires balancing life quality and results.

The i-SDM tool shows treatment side effects, survival chances, and other options clearly. This helps patients think carefully about their choices. Research shows that patients get better at following treatment plans when supported like this.

But how much patients get involved depends a lot on how comfortable they feel with AI and how the healthcare team explains AI advice kindly.

Future Directions in AI Applications for Oncology in the US

Research is still improving AI tools to better serve diverse patients, make them easier to use, and add more features. The CHIEF model is learning to work with rare cancers and to predict reactions to new treatments. Systems like i-SDM want to be easier for older patients who don’t use technology much.

Doctors, patients, and AI developers must keep talking to balance new ideas with practical needs.

Summary Note for Medical Practice Managers and IT Leaders

AI offers new options for oncology clinics in the United States. It helps with diagnosis, treatment planning, patient communication, and workflow. But getting patients and doctors to accept AI is not simple, especially with older adults and in busy clinics. Medical managers should focus on training staff, informing patients, and choosing AI tools that fit well with current oncology work. Some AI companies provide services like phone automation that reduce paperwork while keeping patient support working well.

Putting money into AI solutions that deal with worries about mistaken information, fairness, and ease of use will help cancer clinics improve care quality and work efficiency.