Yatiri Bio’s AML Platform - AI-Driven Agorithms

Yatiri Bio’s AML platform combines deep proteomics with living patient-derived models to map real-world disease biology and drug response. With >250 profiled samples, 40 functional ex vivo models, and over 5,500 dose–response curves across 50 compounds, we connect proteomic signatures to therapeutic outcomes. AI-driven algorithms predict responders and resistance pathways, empowering more brilliant clinical trial design.

Patient-Centered Proteomics

  • Over 100 AML patient samples profiled
  • ~11,000 proteins quantified per sample using high-resolution mass spectrometry
  • Rich metadata includes response, mutation status, and clinical annotations

Living, Testable Patient Models

  • 40 patient-derived ex vivo models built from sorted blasts and clinical samples
  • Captures real AML subtypes and heterogeneity
  • Augmented by public datasets for a broader biological context

Therapeutic Response Mapping

  • >600 drug–cell model combinations tested
  • 5,500+ dose-response curves across 50 compounds (FDA-approved, Phase I, Phase II)
  • Connects proteomic profiles to phenotypic response
  • Predictive Algorithms for Trial Design
  • Deep learning trained on matched proteomic and phenotypic data
  • Identifies predictive response signatures and resistance mechanisms
  • Enables pre-trial patient stratification and therapeutic matching

Clinically Validated and Extensible

  • Developed with Fred Hutch; collaborations pending with Cedars-Sinai, OSU, MD Anderson, and City of Hope
  • Powers internal asset development and partner trials
  • Provides a scalable clinical blueprint for additional indications (e.g., ovarian cancer)

AML Predictive Model