Will AI put me out of a job? Save

The advent of artificial intelligence (AI) continue to amaze us in this changing world and in multiple areas of medicine. As an evolving science, rheumatology has also been introduced to AI where interesting applications seem to enhance patient care.

In the specific case of rheumatoid arthritis (RA), it is known that despite different effective management options, there are still challenges predicting individual patient responses to therapies. The development of AI and other technologies brings new possibilities for optimizing diagnosis, management and has potential for improving outcomes in this disease.

At RheumNow Live 2025, Dr. Jeff Curtis, University of Alabama at Birmingham, presented a talk on, "AI applied to Rheumatoid Arthritis Care: Practically (Not) Perfect". 

How could AI benefit patients with RA?

Large language models (LLMs) such as ChatGPT can synthetize data from signs and symptoms of the disease which can lead to early recognition and prompt patients to seek care earlier.

There is also evidence that these models can provide treat-to-target strategies by analyzing patient individual characteristics and recommendations informed by guidelines and current evidence. A good example is the combination of AI with biomarkers to predict non-response to TNF-inhibitors.

AI also holds potential for predicting RA disease activity which will lead to a better treat-to-target strategy and possibly improving patient outcomes.

How could AI benefit rheumatologists?

It has been no secret that a large burden of clinical practice is documentation and administrative duties. There are different AI models that could potentially reduce burden on clinicians by processing clinical histories in seconds and assisting with note and consult writing, helping to reduce documentation workload. Some LLMs can screen referrals and determine the need to see patients earlier.

Additionally, AI can help physicians to generate a strong differential diagnosis and might assist clinicians in prognosticating outcomes based on clinical data, including comorbidities, disease activity, and physical function.

Does this mean AI will replace us?

While AI takes patient history, summarizes guidelines, recommends best therapeutic options and predict outcomes. It lacks the clinical judgment, empathy, and nuanced decision-making required to manage the complexity of rheumatic diseases which involves a dedicated patient-centered care.

It also has different drawbacks including lack of transparency in how the model works, inconsistency in output (i.e. the LLMs gives you different answers to the same question), “hallucinating” false information, inaccurate prediction of future outcomes among others.

In conclusion, the future of rheumatology will involve collaboration between human expertise and AI-driven insights, improving both patient care and physician efficiency.