Automatic Psoriasis Area and Severity Index (APASI): Artificial Intelligence-Based Quantification for Psoriasis Assessment

Authors

Taig Mac Carthy, Daniel Dagnino, Alfonso Medela, Gonzalo Fernández, Andy Aguilar, Antonio Martorell, Patricia Gómez-Tejerina, Gastón Roustán-Gullón.

Introduction

Psoriasis is a chronic inflammatory skin disease affecting approximately 2-3% of the global population. The Psoriasis Area and Severity Index (PASI) is the gold standard for measuring psoriasis severity in both clinical practice and pharmaceutical trials. However, manual PASI assessment is time-consuming, requires specialized training, and suffers from significant inter-observer variability.

This variability presents a major challenge for clinical trials, where consistent and reproducible measurements are essential for evaluating treatment efficacy. Furthermore, the subjective nature of visual assessment means that the same patient may receive different scores from different clinicians.

Method

We developed APASI (Automatic PASI), an AI-based system that provides fully automated psoriasis severity assessment. The system uses deep learning algorithms trained on thousands of clinical images annotated by expert dermatologists to:

1. Detect and segment psoriatic lesions across body regions
2. Quantify erythema, induration, and desquamation for each lesion
3. Estimate affected body surface area per region
4. Calculate a composite PASI score comparable to expert assessment

The algorithm was validated against a panel of expert dermatologists to ensure clinical accuracy and reliability.

Results

Our results demonstrate that APASI achieves performance comparable to expert dermatologists while significantly reducing inter-observer variability. Key findings include:

• High correlation with expert PASI scores
• Reduced assessment time from minutes to seconds
• Consistent scoring regardless of assessor or location
• Applicability to remote monitoring and decentralized clinical trials

APASI enables objective, reproducible psoriasis assessment that can be deployed in clinical practice, telemedicine consultations, and pharmaceutical research. By removing the subjectivity inherent in manual scoring, APASI has the potential to improve treatment monitoring and accelerate drug development.

Clinical Applications

• Clinical trials: Standardized endpoint assessment across multiple sites
• Telemedicine: Remote severity monitoring for patients with limited access to dermatologists
• Treatment monitoring: Objective tracking of disease progression and treatment response
• Clinical decision support: AI-assisted severity grading to inform treatment decisions

Read full text

Available at https://doi.org/10.1002/jvc2.70143.