By Cary Murray, Frank Fuller, and Eric W. Tramel, Machine Learning Engineers

‍

In our ongoing pursuit of developing novel AI solutions for drug development, we have created digital twin generators (DTGs) for rheumatoid arthritis, plaque psoriasis, and atopic dermatitis. This initiative builds upon our foundational work with Crohn's disease and affirms our dedication to innovation in the crucial areas of inflammation and immunology.

‍

Understanding the Impact

‍

Millions worldwide live with inflammation and immunology conditions, experiencing daily challenges that affect their quality of life. Rheumatoid arthritis, plaque psoriasis, and atopic dermatitis are not just medical terms but realities for many, presenting varied and complex symptoms that significantly impact quality of life. Despite the availability of several treatments, a gap remains in effectively addressing the full spectrum of patient needs.

‍

• Rheumatoid arthritis presents as a systemic inflammatory autoimmune disease characterized by joint stiffness, pain, and swelling. If inadequately managed, it can lead to joint deformities and widespread inflammation throughout the body.
• Plaque psoriasis is a chronic inflammatory skin disorder characterized by thick, scaling red plaques resulting from an unusually high rate of skin cell growth. It usually involves the skin and nails and leads to lesions that cause itching, stinging, and pain. A third of patients with Psoriasis may develop psoriatic arthritis, which is typified by swelling, stiffness, and pain in and around the joints.
• Atopic dermatitis, or eczema, is marked by chronic, dry, itchy rashes, frequently associated with other allergies.

‍

The challenge in treating these conditions lies in their heterogeneity and the complexity of their pathogenesis, often making it difficult to identify universally effective treatments. In rheumatoid arthritis, there's an efficacy ceiling where approximately 30% of patients do not respond adequately to their first biologic or Janus kinase (JAK) inhibitor therapy. In the realm of plaque psoriasis, several effective biologics are on the market, but the lack of generics or biosimilars limits access. The therapeutic advancements for atopic dermatitis, notably with the approval of dupilumab in 2017 and subsequent JAK inhibitor approvals, have markedly changed the treatment landscape for patients with severe conditions. Despite these advancements, issues with efficacy and safety risks persist, underscoring the unmet need for treatments that can fully manage these conditions.

‍

Combinations of existing therapies could improve treatment outcomes, but clinicians' and insurance providers' widespread acceptance and recommendation hinges on robust evidence from clinical trials. Matching patients with the most effective treatment, possibly through biomarker-driven precision medicine, could enhance treatment efficacy. Conducting comparative analyses between different therapeutic strategies could also expedite patient relief by providing clear evidence to guide treatment selection.

‍

Nevertheless, the advancement of these innovative treatment strategies faces substantial hurdles. The potential high costs of diagnostic testing and the increased demand for trial participants often lead to bypassing these important investigations. This situation underscores the critical need for continuous innovation beyond the current therapeutic offerings and the integration of precision medicine approaches to fully address the complexities of these conditions.

‍

The power of Neural Boltzmann Machines in DTGs

‍

Unlearn’s DTGs are powered by Neural Boltzmann Machines (NBM), a new kind of AI model we invented that can learn and model the complexity of real patients' biological data.

‍

Learning from Data: The NBM is first trained on past clinical data. The model learns the intricate relationships between different health variables and how they contribute to the outcomes of interest, such as disease progression or treatment response.

‍

Understanding Individual Variability: The NBM is able to capture how subtle variations in clinical assessments, demographics and laboratory results correlate to the time progression of various disease indicators. These correlations are crucial for generating accurate digital twins that reflect the unique health profiles of individual patients.

‍

Generating Digital Twins: Once trained, the NBM can generate digital twins by simulating the health trajectories of new patients based on their initial health data and the patterns learned during training. Given a new patient's baseline health information, the NBM uses its learned knowledge to predict how this patient's health might evolve. 

‍

Indication-Specific Innovations

‍

Each DTG is designed to address the unique challenges and trial designs of its respective indication:

‍

• The Rheumatoid Arthritis DTG supports commonly measured outcomes used in treatment-to-target treatment designs like American College of Rheumatology Criteria 20 (ACR-20), Disease Activity Score-28 (DAS-28), Clinical Disease Activity Index (CDAI), and Simplified Disease Activity Index (SDAI). 
• The Plaque Psoriasis DTG supports primary continuous and binary outcomes based on the Psoriasis Area and Severity Index (PASI), Investigator Global Assessment (IGA), and Body Surface Area (BSA). It can predict progressions for individuals who have recently initiated therapies to create digital twins for use in an active comparator arm. 
• The Atopic Dermatitis DTG forecasts continuous and binary endpoints derived from the Eczema Area Severity Index (EASI), Scoring Atopic Dermatitis (SCORAD), and Investigator Global Assessment (IGA).

‍

TwinRCTs: Quick and confident clinical trials

‍

Our TwinRCT™ leverages participants’ digital twins to improve the ability to observe treatment effects in early-stage studies by increasing power without adding more patients. For late-stage studies, TwinRCTs provide the advantage of reaching full enrollment sooner because they require fewer participants to achieve the same power as traditional trial designs.

Our method using participants’ digital twins in clinical trials was qualified by the European Medicines Agency and adheres to current FDA guidance. These endorsements underscore our commitment to providing scientifically robust and regulatory-compliant solutions in drug development.

‍

Join us in transforming clinical trials

‍

The introduction of these DTGs represents a significant milestone in the journey to combat inflammatory diseases. We invite industry leaders, researchers, and clinicians to join us in this innovative phase of clinical research, where innovation meets efficacy to bring about a healthier future for all.

We encourage you to download our DTG specification sheets to gain a comprehensive understanding of our DTGs, including their capabilities and underlying data.