Drug Simulation Applications of Gemini Digital Twins


Aitia partners with leading biopharma companies to simulate their drug candidates across our cohort of Digital Twins in specific diseases to prioritize drug targets, discover optimal combination therapies, and discover responding versus non-responding patient groups and corresponding biomarkers. Aitia brings the full power of the world’s most accurate Digital Twins to help advance and de-risk promising drug candidates. Current Gemini Digital Twins are used in multiple myeloma, prostate cancer, Alzheimer’s Disease, Parkinson’s Disease, and Huntington’s Disease.




Gemini Digital Twin Key Functionalities

Discovery &

  • Prognostic Marker Discovery
  • Combo Therapies


  • Prognostic-Novel Drivers of
  • Precision Patient Stratification for
    I/E criteria
  • Response Markers/Sub-Population
    (SoC Drugs)
  • In Silico Head-to-Head trial
  • Response Markers/Sub-Population
    (drug candidate in RCT)
  • Historical Control Arms

Fueling Translational Research

Unmet Need: Lack of strong predictive markers beyond MMSET for stratifying MM patients with high-risk of progression, hampering clinical trial design and optimal patient care

Insights: Causal Network simulations linked patient characteristics and gene expressions to novel biological pathways and clinical outcomes, revealing PHF19 as a new marker of high-risk disease

Impact: Established PHF19 as a stronger predictor of MM progression than the conventional high-risk marker MMSET, enabling better designed clinical trials that better achieve clinical endpoints. Additionally, PHF19 is now part of a 4-factor predicitive model that can effectively stratify patients in clinical settings

Unmet Need: Limited understanding of the mechanistic pathways around Androgen Receptor (AR) for developing next gen treatments for castrate sensitive and castrate resistant prostate cancer and associated prognostic biomarkers of overall survival

Insights: Gemini’s hypothesis free approach revealed novel germline mutations that modulate the effect of Androgen Receptor (AR) Copy Number Gain and effects on AR gene overexpression in metastatic Castration-Resistant Prostate Cancer (mCRPC)

Impact: This causal approach reveals new potential targets as well as novel prognostic markers of survival. These findings will help both early research discovery efforts for new target discoveries and translational endeavors to design sophisticated clinical trials and simulations of control and efficacy arms

Forest with fog

Accelerating Clinical Trials

Unmet Need: Sidedness of colorectal tumors has been used as a predictor of prognosis and response to standard of care drugs Erbitux and Avastin

Insights: Gemini models fueled with DNA sequence and gene expression and clinical data have revealed the underlying molecular circuitry of “sidedness”. Simulations of these models reveal novel prognostic molecular drivers of differential response between left and right-sided tumors, and the subpopulations that achieve a survival benefit of 11 months when treated with Avatsin vs Eribitux

Impact: Better treatment of mCRC patients as it revealed that at least 5% of patients are likely being mistreated per current guidelines. These models are also utilized to better determine inclusion/exclusion criteria for mCRC trials

Unmet Need: There is an inability to determine ahead of treatment which patients will respond or not respond to stem cell transplant, an invasive and expensive procedure with serious side effects

Insights: Simulation of the causal network models revealed the CHEK1 gene expression levels, along with 2 other supporting genes, as strong predictors of response/non-response conveying ~20 months of survival benefit. An independent trial at Dana Farber Cancer Center validated the survival benefit of ~20 months

Impact: If further validation of this predictive marker is successful, this development would be transformative for patient care by allowing patients to avoid an invasive therapy that would be ineffective and potentially toxic to them, vs. earlier treatment with treatment that would be effective for them. This insight would also allow for better stratified clinical trials for therapies that could be used in early lines of treatment

Unmet Need: Parkinson’s disease progression is highly variable across patients, making it challenging to treat and design effective clinical trials. We aimed to discover the unknown molecular and genetic drivers of motor progression to design more efficient clinical trials

Insights: Parkinson’s disease models built from clinical, molecular, and genetic data confirmed known predictors and identified novel predictors of Parkinson’s motor progression, including the biological context of potential progression markers. Specifically, patients who carried minor alleles of both rs17710829 (in LINGO2; Leucine-rich protein, like LRRK2) and rs929897 (nearest gene is DPP10, seen in Alzheimer’s Disease) SNPs had substantially faster rate of motor decline. Findings were validated through an independent patient cohort and a method to prospectively differentiate rates of motor progression was discovered

Impact: Discoveries could reduce trial enrollment by 20%. Additionally, findings provided insight into the mechanisms of disease process, leading to potential for novel therapeutic intervention and improving patient care by aiding in clinical disease management. Results were published in high-impact journal Lancet Neurology (see link below)

Clinical Trials