top of page
The Criteria of
Virtual Patients = Human Patients
for determining drug clinical efficacy
99%
CONFIDENCE
>118
SAMPLE SIZE
>75%
ACCURACY
G O L D
STANDARD
118 public prospective predictions of clinical trial outcomes is the minimum sample size and 75% is the minimum prediction accuracy to show virtual patients can substitute for human patients in clinical trials with 99% Confidence Interval (CI).
All major disease trials represent one of the most challenging tests of building virtual patients that are CLINICALLY EQUIVALENT to human patients.
A virtual patient is supposed to be a general-purpose platform that could functionally reconstruct all major diseases that could possibly occur to a single human patient.
Our all major disease virtual patients have achieved an 22-month-long public track record of 84.4% (130/154) accuracy in our prospective predictions of phase 2 and phase 3 clinical trial results.
84.4%
PREDICTION ACCURACY
as of 03 December 2021
for all therapeutic areas excluding genetic disorders and hematological disorders.
First-in-class trials represent one of the most challenging tests of building virtual patients with ZERO prior clinical data on a single drug-indication pair.
80%
VALIDATION TRIALS
Our first-in-class virtual patients has achieved an 22-month-long public track record of 84% (103/123) accuracy in our prospective predictions of first-in-class phase 2 and phase 3 clinical trial results.
84%
PREDICTION ACCURACY
as of 03 December 2021
for all therapeutic areas excluding genetic disorders and hematological disorders.
80% validation trials (123/154) are first-in-class pivotal phase 2 and phase 3 clinical trials for specific target-indication pairs.
Immuno-oncology represent one of the most challenging tests of building virtual patients with MOST COMPLEX trial designs featuring active controls and combo therapies.
84%
PREDICTION ACCURACY
Immuno-oncology is also one of the most challenging tests of building virtual patients with active controls and combo therapies as the standard design of pivotal clinical trials.
In additions, human cancer patients are further stratified by certain biomarkers, stages of cancer and prior treatment history, all of which add to the complexity of building virtual cancer patients.
Our immuno-oncology virtual patients have achieved 84% accuracy (26/31) in predicting the clinical efficacy of immuno-oncology pivotal phase 2 and phase 3 clinical trials, including trials with the most complex design protocols featuring multiple-arms and multiple co-primary endpoints (e.g. CheckMate 648).
Alzheimer's Disease represents one of the most challenging tests of building virtual patients with
ZERO POSITIVE prior clinical data on all tried drugs.
100%
TRUE POSITIVE PREDICTION
0%
FALSE POSITIVE PREDICTION
Alzheimer's Disease is one of the most challenging tests of building virtual patients with ZERO POSITIVE prior clinical data about disease-modifying first-in-class drugs.
The gold standard of validating AD virtual patients is 0% false positive and 100% true positive when it comes to determine whether first-in-class AD drug that can SUCCEED in well-designed phase 3 pivotal clinical trials.
Our AD virtual patients have achieved the gold standard by having predicted failure for almost all AD trials (0% false positive rate) and having predicted success for masitinib, a first-in-class c-Kit inhibitor that met the efficacy primary endpoint in the pivotal phase 3 trial, the first worldwide in the past 20 years (100% true positive rate).
as of 03 December 2021
85.4%
PREDICTION ACCURACY
COVID-19 is one of the most challenging tests of building virtual patients with ABSOLUTELY ZERO prior clinical data and SHORTEST amount of time.
Demiurge built and published the world's first COVID-19 virtual patient in March 2020, just 6 weeks since WHO confirmed the human-to-human transmission of SARS-CoV-2.
Our COVID-19 virtual patients have achieved 85.4% accuracy (35/41) in predicting the clinical efficacy of 50+ first-in-class COVID-19 vaccines and therapeutics, in a 22-month public track record of prospective outcome prediction of COVID-19 pivotal phase 2 and phase 3 trials.
6
WEEKs
COVID-19 represents one of the most challenging tests of building virtual patients ASAP with ABSOLUTELY ZERO prior clinical data.
as of 03 December 2021
Discover the first principles of neural networks and the causal pathophysiology of complex diseases by bridging deep learning and life science.
Demiurge has achieved an 22-month-long public track record of 84.4% (130/154) accuracy in our prospective predictions of phase 2 and phase 3 clinical trial results.
84.4%
PREDICTION ACCURACY
as of 03 December 2021
for all therapeutic areas excluding genetic disorders and hematological disorders.
All Diseases
123
POTENTIAL FIRST-IN-CLASS TARGETS
31
POTENTIAL BEST-IN-CLASS TARGETS
80% validation trials (123/154) are first-in-class pivotal phase 2 and phase 3 clinical trials for specific target-indication pairs.
Demiurge has achieved the world's longest 22-month-long public track record of 84% (103/123) accuracy in our prospective predictions of first-in-class phase 2 and phase 3 clinical trial results.
as of 03 December 2021
First-in-class targets
PREDICTION ACCURACY
84%
Demiurge has achieved an 22-month-long public track record of 84% (103/123) accuracy in our prospective predictions of first-in-class phase 2 and phase 3 clinical trial results.
as of 03 December 2021
for all therapeutic areas excluding genetic disorders and hematological disorders.
First-in-class trials
Demiurge has achieved an 22-month-long public track record of 84% (26/31) accuracy in our prospective predictions of immuno-oncology phase 2 and phase 3 clinical trial results.
84%
PREDICTION ACCURACY
as of 03 December 2021
Immuno-oncology trials
as of 03 December 2021
TRUE POSITIVE RATE
100%
Our AD virtual patient has achieved the gold standard by having predicted failure for almost all AD trials (0% false positive rate) and having predicted success for masitinib, a first-in-class c-Kit inhibitor that met the efficacy primary endpoint in the pivotal phase 3 trial, the first worldwide in the past 20 years (100% true positive rate).
FALSE POSITIVE RATE
0%
AD Trials
Even for new diseases like COVID-19 without any preexisting clinical data, Demiurge published a detailed COVID-19 disease model in March 2020 (Lovetrue, 2020 here).
Demiurge has achieved an 18-month-long public track record 85.4% (35/41) accuracy in prospective predictions of phase 2 and phase 3 COVID-19 clinical trial results in 2020/2021.
PREDICTION ACCURACY RATE
85.4%
as of 03 December 2021
COVID-19 Trials
All Indications
130+
INDICATIONS
THERAPEUTIC AREAS
26
Demiurge has made 280+ prospective predictions in public covering more than 130 indications in 26 therapeutic areas, including oncology, Central Nervous System, cardiovascular, musculoskeletal, immunomodulators, diabetes, metabolic, infectious diseases, dermatology, seneory organs, endocrine, genito-urinary etc.
as of 03 December 2021
bottom of page