Why Pheiron and Why Now?

Drug development is broken: Over 90% of drug development programs entering clinical development fail, costing the industry an average of $6.16bn for a novel approved drug1. The root of the problem? Insufficient evidence to anticipate clinical efficacy. We build Pheiron to change that - in silico and at scale.

We need better evidence

The pharmaceutical industry is suffering from high failure rates and spiraling R&D costs. An estimated 50% of clinical trials fail due to a lack of clinical efficacy2,3. Establishing clinical efficacy, has two essential prerequisites: the right target for the disease and the right population for the trial. 

Genetic evidence on drug targets drawn from millions of individuals has revolutionized the early phases of target identification and validation. Drug programs backed by genetic evidence have been found to be >2x more likely to go past approval and overall, programs supported by genetic evidence have an 80% higher chance to advance to clinical development4–6.

Once a target has been identified and a drug candidate designed, demonstration of clinical efficacy in a randomized controlled trial is the crucial milestone for approval. However, even when drugs show potential in early-stage research or preclinical settings, clinical trials routinely fail. Matching the pathology of the trial population with the MoA of the drug candidate has been shown to increase the likelihood of trial success by 1.5-8x6. Prognostic enrichment of trial populations for example, allowed early statin trials to demonstrate efficacy in only seven (!) individuals with familial hypercholesterolaemia7

A lot of effort in drug development revolves around streamlining processes or reducing overall cost. However, improving the overall quality of program decisions is much more important: reducing the Phase-II failure rate by 20% corresponds to an average 450mn dollars saved per program8. The quality of program decisions fundamentally depends on the ability to anticipate clinical utility in real humans. Basing decisions on the right evidence is critical for success9,10.

We need better evidence to navigate the road to success.

Clinical outcomes are key

Evidence from population-scale data can inform on target safety and efficacy before entering clinical development. And in clinical development, evidence from population-scale data can inform biomarker discovery and patient selection. Building evidence on clinical efficacy, however, requires a deep understanding of clinical care. 

Currently, we are witnessing a cambrian explosion of biomedical data. Broad and scalable -omics assays and high-resolution imaging record phenotypic information in unprecedented breadth and depth. This deep phenotypic data contains information on the disease pathology. At the same time, longitudinal health records represent individual health trajectories as a time-resolved collection of a patient’s medical history. Health records capture information on clinical outcomes.

Unfortunately, to date, high-dimensional biomedical data is disconnected from the clinical realities and the real-world outcomes which patients experience, leading to a high uncertainty in understanding disease on a phenotypic level. This touches the core of medicine itself. To date, we still cannot answer the most basic medical questions for the vast majority of diseases: who gets sick in the first place? What characterizes people once they are diagnosed? Who continues to get complications? 

Connecting the disease pathology recorded in deep phenotypic data with real-world clinical outcomes is the missing answer to the clinical questions. And it is the key to building predictive evidence which anticipates clinical efficacy. 

This is personal to us

We deeply care about this evidence. We have experienced the consequences of the missing link first hand. As clinicians we cared for patients and faced the harsh reality of our limited medical understanding: not being able to provide the best care for our patients.

As scientists, we glimpsed at a world, in which at least a few of the fundamental questions of medicine were answered. We demonstrated that we could use AI to bridge from complex biomedical data to clinical outcomes11, to integrate genetics12, and to scale risk modeling over thousands of diseases simultaneously13. We leveraged population-scale cohorts and navigated the information maze of health records.

Now, as builders we aim to answer them at scale. That's what Pheiron is all about. 


With pheiron we are on a mission to provide predictive evidence necessary to develop safe and effective drugs - in silico and at scale.

Our platform integrates billions of data points from complex biomedical data across population cohorts. We use AI to systematically link deep phenotypic data with real-world health trajectories, bridging from detailed disease pathology to clinical outcomes. We have identified thousands of disease signatures from more than 30 complex data modalities including 12-lead ECGs, spirometries, cardiac MRIs, retinal fundus photographs, broad -omics assays (metabolomics, proteomics, transcriptomics) and established blood counts and biochemistry. Importantly, integrating information from deep phenotypic data allows us not only to assess evidence across layers of biology, but also to build evidence for indications with notoriously elusive phenotypes. Our platform helps program centered teams to systematically de-risk their decision making, build conviction and improve their likelihood of success. Customers use our platform in two key areas: evidence for target safety and efficacy and evidence for selecting the right trial population. Our disease signatures allow us to pick up even subtle changes in phenotypic information, boosting the quality of our generated evidence. Using our disease signatures as prognostic markers, trials can be optimized to reduce failure risk and cost by increasing statistical power while reducing the required population size. 

In an ever more competitive industry deeply dependent on positive R&D output, the right evidence can be decisive. Our platform helps drug developers to sharpen their edge and focus on what they do best: develop new drugs.

If the problems we are solving resonate with you and your team, please reach out to info@pheiron.com. We would love to help.

The Best is Yet to Come

We are just at the very beginning. We know drug development and healthcare are constantly evolving. The target is moving and we will continue to partner with leading enterprises and scale our world-class team to become the foundational insights layer on phenotypic information. 

Again, if you find this exciting, want to contribute or exchange thoughts, reach out to info@pheiron.com. Let’s talk!


1. Schuhmacher, A., Hinder, M., von Stegmann Und Stein, A., Hartl, D. & Gassmann, O. Analysis of pharma R&D productivity - a new perspective needed. Drug Discov. Today 28, 103726 (2023).

2. Harrison, R. K. Phase II and phase III failures: 2013-2015. Nat. Rev. Drug Discov. 15, 817–818 (2016).

3. Dowden, H. & Munro, J. Trends in clinical success rates and therapeutic focus. Nat. Rev. Drug Discov. 18, 495–496 (2019).

4. Nelson, M. R. et al. The support of human genetic evidence for approved drug indications. Nat. Genet. 47, 856–860 (2015).

5. Minikel, E. V., Painter, J. L., Dong, C. C. & Nelson, M. R. Refining the impact of genetic evidence on clinical success. bioRxiv (2023) doi:10.1101/2023.06.23.23291765.

6. Wong, C. H., Siah, K. W. & Lo, A. W. Estimation of clinical trial success rates and related parameters. Biostatistics 20, 273–286 (2019).

7. Mabuchi, H. et al. Effects of an inhibitor of 3-hydroxy-3-methylglutaryl coenzyme A reductase on serum lipoproteins and ubiquinone-10-levels in patients with familial hypercholesterolemia. N. Engl. J. Med. 305, 478–482 (1981).

8. Bender, A. & Cortés-Ciriano, I. Artificial intelligence in drug discovery: what is realistic, what are illusions? Part 1: Ways to make an impact, and why we are not there yet. Drug Discov. Today 26, 511–524 (2021).

9. Scannell, J. W. & Bosley, J. When Quality Beats Quantity: Decision Theory, Drug Discovery, and the Reproducibility Crisis. PLoS One 11, e0147215 (2016).

10. Scannell, J. W. et al. Predictive validity in drug discovery: what it is, why it matters and how to improve it. Nat. Rev. Drug Discov. 21, 915–931 (2022).

11. Buergel, T. et al. Metabolomic profiles predict individual multidisease outcomes. Nat. Med. 28, 2309–2320 (2022).

12. Steinfeldt, J. et al. Neural network-based integration of polygenic and clinical information: development and validation of a prediction model for 10-year risk of major adverse cardiac events in the UK Biobank cohort. The Lancet Digital Health 4, e84–e94 (2022).

13. Steinfeldt, J. et al. Medical history predicts phenome-wide disease onset and enables the rapid response to emerging health threats. bioRxiv (2023) doi:10.1101/2023.03.10.23286918.

Align between targets, assets and populations.

Subscribe to our blog and stay tuned on Predictive Evidence in drug development.