shatter transparency strip-07.png

AI-Driven Patient-Centric Drug Discovery

Improving the chances of successful clinical translation

The current drug discovery process is long, costly, and suffers from high attrition rates due to poor translation of discoveries into the clinic. Much of this lack of success can be traced back to the early part of discovery which accounts for about one-third of the total cost and takes about 6 years. 

 

Leveraging our core competence in profiling the drug response of patient tumors, our large knowledgebase of tumor drug response and other data, together with proven AI, Helomics has created a unique capability for oncology drug discovery that allows for the highly efficient screening of drug responses from thousands of diverse, well-characterized patient primary tumor cell lines.

This novel disruptive patient-centric approach is ideally suited to the early part of drug discovery (especially Hit-to-lead, lead optimization, and pre-clinical), resulting in better prioritization of compounds and better coverage of patient diversity.  This will dramatically improve the chances of successfully translating discoveries into the clinic, resulting in lowered costs, shortened timelines, and most importantly enhanced "speed-to-patient" for new therapies

Addressing the key challenges in oncology drug discovery

 

Exhaustive screening across thousands of heterogeneous tumors with existing technologies is cost and time prohibitive.

Assessing a combination of drugs in the context of heterogeneity is even more prohibitive.

The key goal of early discovery is to deliver a drug candidate into clinical trials that is both effective and safe, yet researchers often fall short of that goal with over 80% of compounds failing to translate into the clinic because they lack efficacy or have toxicity issues. For oncology discovery, many of these failures result from inadequately addressing the problem of heterogeneity, both within and between tumors  - the tumor is a complex ecosystem of different cell types,  and a tumor from Patient A is not the same as the one from Patient B.

Furthermore, real patient sample testing typically occurs in late-stage development using expensive and time-consuming xenograft or other PDx platforms.

Our solution is to use our AI-driven patient-centric drug discovery platform to efficiently screen sets of compounds over thousands of highly characterized patient tumor primary cell lines and provide

  • Results on the diversity of drug responses across a set of patients,

  • Correlations and comparisons with existing  standard-of-care drugs

  • A model of  different drug responses for patients with different tumor properties

  • Results of the responses of patients to drug combinations if required

Introducing PeDAL - Patient-centric Drug Discovery using Active Learning

PeDAL is a unique technology that combines a proprietary, clinically validated patient tumor cell line assay, a vast knowledgebase of proprietary and public data together with active learning - the active learning allowing the efficient exploration of compound drug responses against a large diverse patient "space". PDAL offers researchers the opportunity to efficiently and cost-effectively bring patient diversity into drug discovery much earlier.

TruTumor 

Clinically validated (ovarian cancer), proprietary patient primary tumor culture assay

 

1000's of patient tumor cell lines

Read more ->
Knowledgebase

Proprietary TumorSpace model of 150,000 drug response profiles from clinical testing

Proprietary model of all public drug-target studies

Read more ->
CoRE AI

Proven CoRE active learning approach for constructing predictive models of drug response to efficiently guide multiple rounds of drug response testing using TruTumor

Read more ->
How does PeDAL work?

PeDAL works by iterative cycles of active-learning powered Learn-Predict-Test to guide the testing of patient-specific compound responses using the TruTumor assay and patient cell lines to build a comprehensive predictive model of patient responses to compounds. This predictive model can then be used to rank compounds by the fraction of patients of certain profiles that respond as well as the set of compounds that provide the best coverage across patients.

© 2020 Helomics Corp. All rights reserved.

  • White Twitter Icon
  • Facebook - White Circle
  • White LinkedIn Icon