Modeling and AI

The LANCE platform to identify novel therapeutic immunometabolic pathways

About AI

AI techniques and computational modeling can be used to further our knowledge about immunology and drug development for infectious and autoimmune diseases. Utilizing these capabilities allow us to run in silico experiments to create predictions for in vitro and in vivo results. Simulators of synthetic populations and ODE models are examples of powerful tools that allow us to study biological processes involved in immunology and how these processes can be targeted and manipulated by new, novel therapeutics.


The NIMML Institute focuses on using AI techniques and computational modeling to further its position as a pioneer in the field of immunometabolism. A pipeline has been developed and utilized by NIMML to identify key connections between immunology and metabolism and target them effectively with potential drugs. By utilizing these techniques, NIMML stays on the cutting edge of immunometabolic research and drug development for immunoregulatory control and autoimmune therapy.

AI pipeline allows for simulation of virtual treatments to populations after experimental data collection

Applications of AI

The NIMML Institute has developed an in silico pipeline to predict clinical efficacy of innovative C. difficile treatments with nonclinical results. Utilizing supervised machine learning, simulated output quantities (i.e. time of clearance, quantity of commensal bacteria, T cell ratios) were transformed into clinical predictions based on linear and nonlinear correlations between the output quantities and prior clinical trial data. The NIMML Institute is on the forefront of developing and employing methods that reduce material and time costs. AI is used to guide in-house biological experiments for more precise and useful results.
The NIMML Institute has explored multiple methodologies to determine the efficacy of various advanced machine learning and artificial intelligence (AI) techniques for modeling CD4+ T cell differentiation, a key immunological process that is non-linear, heterogeneous and highly complex. We compared three supervised learning methods for modeling CD4+ T cell differentiation: Artificial Neural Networks (ANN), Support Vector Machines (SVM), and Linear Regression (LR). Application of supervised machine learning methods has the potential reduce immunological complexity by focusing on the input and output cytokines during CD4+ T cell differentiation. In addition, this modeling framework can be efficiently integrated into multiscale tissue-level models of the immune system. The NIMML Institute stays on the cutting edge of computational immunology by continuously optimizing AI-based and computational modeling-based methodologies for fast and accurate predictions from its models.

The NIMML Institute applied AI-based methods to develop a unique pipeline that produces useful insights for testing the clinical efficacy of IBD therapeutics in a randomized in silico trial. Crohn’s disease (CD) and ulcerative colitis (UC) are the two clinical manifestations of IBD, a widespread and debilitating autoimmune disease of unknown etiology afflicting 5 million people worlwide. Current therapies for IBD include the use of corticosteroids, antibiotics, immune-modulators, and anti-TNF-α biologics with limited efficacy and significant side effects. Thus, there is an unmet clinical need for safer and more effective oral therapeutics. To develop path to safer cures, we designed and implemented a novel integrated approach, based on the application advanced machine learning and AI algorithms. We created a large synthetic cohort of CD and UC patients and engineered these populations using immunological variables data derived from published clinical trials. For the proposed Phase III placebo-controlled, randomized in silico clinical trial, we simulated a test set of synthetic patients for five therapeutics. We randomly assigned five treatments -nutritional intervention (conjugated linoleic acid), a Phase II therapeutic (GED-0301 -Mongersen), a novel pre-clinical therapeutic targeting lanthionine synthetase C-like 2 (LANCL2), an anti-Tumor necrosis factor alpha (anti-TNF-α) antibody, and placebo to each synthetic patient. In CD trials, the change in Crohn’s disease activity index (CDAI) post 6 weeks was used to evaluate the effectiveness of current or investigational CD therapeutics. The change in CDAI score was based on the change in levels of biomarkers TNF-α (tumor necrosis factor-alpha) and IFN-γ (interferon gamma), which differed for every therapeutic intervention. For CD, the in silico study results demonstrated that therapeutics targeting LANCL2, ameliorated clinical disease in severe CD patients, with an average drop of 127 points from the initial CDAI score. The effectiveness of the LANCL2 therapeutics was comparable with anti-TNF-α antibodies. The analysis also highlighted that young patients (age < 40 years) respond significantly (P < 0.0001) better to treatment than older patients (age> 40 years).The study highlighted the value of in silico clinical trials for acceleration and improvement of drug development process. It sets a data-driven path for prediction of clinical outcome of investigational therapeutics forr autoimmune disease that leverages the advancements in machine learning and AI. The study opens up prospects that can aid in the future design of targeted precision medicine interventions combining biomarkers and therapeutic intervention for IBD patients with a long-term benefit of informed clinical development plans.