CURATE.AI | AI Shows Promise for Supporting Rare Blood Cancer Treatment

CURATE.AI, an AI tool developed by the National University of Singapore’s School of Medicine (NUS Medicine, SG), has demonstrated significant enhancement in treatment outcomes for a rare type of blood cancer.

In a single trial, the AI-powered clinical decision support system CURATE.AI was utilized to calibrate treatment plans based on a single individual’s data. Published in the NPJ Digital Medicine journal, the study departs from large dataset analysis, focusing on individual patient information to offer a novel approach to personalized treatment delivery.

Methodology and Results

Over a two-year period, CURATE.AI analysed the data and outlined a treatment plan for a patient suffering from Waldenström Macroglobulinemia (WM), a rare blood cell cancer characterized by an excess of white blood cells in the bone marrow.

CURATE.AI operates in different phases: a calibration phase where physicians propose drug doses within a safety range followed by an efficacy-driven dosing phase where it recommends optimal doses based on the patient’s profile.

The system generates and updates a personalized profile using dose-response data pairs, with physicians actively involved in building the small dataset used for dose recommendations.

The process allowed recalibration when patients experienced systemic changes, and drug doses were adjusted weekly within the standard-of-care dose range.

During the efficacy-driven dosing phase, CURATE.AI was considered relevant in more than 88% of treatment cycles, and its clinical advice was accepted for prescription in 95.7% of those cases.

The treatment resulted in significant benefits for the patient. Red blood cell levels improved markedly, eliminating the need for blood transfusions. Additionally, the patient experienced no severe adverse reactions to the treatment.

These results highlight CURATE.AI’s potential ability to outline effective personalised treatment plans that improve patient therapy outcomes.

Professor Dean Ho, study co-lead and NUS Medicine’s Director of the Institute for Digital Medicine (WisDM) stated:

“Our study highlights the effectiveness of using small data to treat extremely rare diseases— addressing the gaps where traditional big data methods fall short, and where largescale trials are not feasible due to the limited patient population.”

Why This Matters

CURATE.AI’s initiative is in line with Project Optimus, an FDA-led initiative to update cancer dose optimization standards.

The initiative promotes determining dosage based on clinical and nonclinical data to reduce side effects and optimize benefits for patients.

The application of AI-powered tools to treat rare conditions such as WM faces significant challenges, especially due to the limited data available to effectively design therapy plans for uncommon diseases.

In response to this concern, the NUS Medicine study showcases AI’s potential to complement and improve human clinical efficacy for rare disease treatment.

The study suggests that AI-based tools such as CURATE.AI could potentially revolutionize rare disease treatment by monitoring patients’ data, suggesting tailored treatment adjustments and predicting potential complications before health conditions worsen.

“No two patients are alike, and even the same patient can change over time as well. It is essential for treatment to evolve alongside the patient,” stated Professor Dean Ho.

In the future, this novel approach to personalized care could reduce the time invested in the experimental investigation of rare disease therapy, potentially saving lives by providing more appropriate treatment adapted to each patient’s unique disease development.

Near-future Challenges and Limitations

Despite AI’s recent and promising contributions to advances in different areas of healthcare including cancer, vaccine production and depression, the implementation of AI in rare disease therapy is still in its early stages.

Whilst promising, the results showcased by the NUS study are based on a single isolated case study, and further investigation across a wider patient population will be needed to corroborate its findings.

Additionally, issues involving patient data protection and current AI device reliability need to be thoroughly investigated and addressed by future studies.

Before the application of AI in healthcare can be considered reliable, further research is needed to address concerns involving ethical standards and regulatory requirements that might compromise the interests of patients.