New AI Framework Enables Earlier Parkinson's Disease Detection Through Questionnaire Analysis

Quest2Dx AI Framework Transforms Parkinson's Disease Screening

Scientists at Rensselaer Polytechnic Institute have developed Quest2Dx, an AI framework designed to improve early screening for Parkinson's disease by analyzing lengthy health questionnaires more accurately than existing models1

1

. Published in Nature Aging, the research addresses a critical gap in healthcare: identifying potential Parkinson's cases years before physical symptoms become apparent2

2

.

Led by Pingkun Yan, Head of the Department of Biomedical Engineering at Rensselaer Polytechnic Institute, the team recognized that current diagnostic approaches rely heavily on observation of motor symptoms and invasive procedures like spinal taps. Yet Parkinson's disease may be present for years before these physical manifestations emerge. "People with Parkinson's disease get referred to specialists pretty late," Yan explained. "We built this AI model to screen for potential cases earlier"1

1

.

Addressing Technical Challenges in Health Questionnaires

The Quest2Dx deep learning framework tackles three key technical challenges that plague conventional AI models for Parkinson's screening. It handles missing responses without relying on traditional imputation methods, works across multiple questionnaires with different wording even when trained on just one, and calculates the most relevant question-answer pairs for each individual patient2

2

.

Primary care clinics frequently deploy health questionnaires for disease screening, but these tools present unique challenges. The Parkinson's Precision Medicine Initiative database provided a questionnaire with 187 questions for training, while the Fox Insight Study validation dataset contained more than 300 items1

1

. Patients frequently skip questions in such lengthy assessments, creating gaps that traditional machine learning algorithms struggle to handle effectively.

Conventional models attempt to fill these blanks through imputation, a process that introduces errors and obscures hidden patterns. These systems also fail when questions are reworded or reordered because they learn positional data rather than semantic meaning2

2

.

Large Language Models Enable Semantic Understanding

The research team, including Director of the Center for Biotechnology and Interdisciplinary Studies Juergen Hahn and former doctoral student Diego Machado Reyes, leveraged large language models to overcome these limitations. "We used the large language model to understand the meaning and context of each question and answer," Yan said. "We used the AI model to read those questions, to read the answers, and then classify whether the patient may or may not have Parkinson's disease"1

1

.

This semantic approach allows Quest2Dx to differentiate between patients with Parkinson's symptoms and healthy control groups more effectively than earlier frameworks. Analysis showed the AI model for Parkinson's screening outperforms previous approaches, including systems built on the pre-trained large language model Llama 22

2

.

Identifying Subtle Early Warning Signs

The early detection of Parkinson's disease hinges on recognizing subtle symptoms that patients and caregivers often dismiss as minor issues. These include loss of smell, sleep disorders, handwriting difficulties, and mood changes—signs that can precede motor symptoms by years1

1

. Such delays in recognizing warning signs push back neurologist referrals, potentially missing crucial windows for intervention.

Quest2Dx identified the most informative questions as those addressing difficulty with common tasks like getting out of bed and eating, alongside risk factors such as exposure to industrial solvents and degreasers. The system also flagged REM sleep behavior disorder, anxiety, and elevated feelings of tension or stress as potential predictors2

2

. Because these conditions overlap with many other illnesses, no single answer provides diagnostic certainty on its own. However, when considered together—such as solvent exposure alongside other early signs—they become clinically relevant1

1

.

Implications for Healthcare Delivery

The development carries significant implications for primary care settings where comprehensive questionnaires are standard but often underutilized due to their complexity. Healthcare providers may overlook subtle Parkinson's-related signs buried within lengthy assessments. By automating this analysis, Quest2Dx enables more efficient screening without adding burden to already-stretched clinical workflows.

The team's collaboration with University of Pennsylvania bioinformatics professor Li Shen and support from the Michael J. Fox Foundation for Parkinson's Research through the Parkinson's Precision Medicine Initiative demonstrates the cross-institutional effort required to advance early screening capabilities2

2

. Watch for potential clinical trials testing Quest2Dx in real-world primary care environments, which would provide crucial data on its practical effectiveness and integration challenges. The ability to screen patients years before motor symptoms emerge could reshape treatment timelines and improve long-term outcomes for those at risk of developing Parkinson's disease.