How MBRS issues biological risk scores
This page is a high-level description of the framework used to produce issued, versioned outputs. Detailed weightings and proprietary implementation are intentionally omitted.
MBRS Bio is a computed system. Outputs are generated from mechanistic modeling and quantitative pipelines—not from literature lookups or text-based summarization. Any alignment with known findings is incidental and serves as validation.
For internal research and licensed analytical use only. Not medical advice, diagnosis, or regulatory guidance.
Key metrics
Short definitions (0–1 unless noted)| Headline | Discrete conclusion label summarizing overall net balance of efficacy vs risk signals (e.g., favorable / mixed / unfavorable), derived from the combined score stack. |
| Bio-Score | Primary continuous biological risk score summarizing the full evidence stack into a single standardized value suitable for cross-program comparison. |
| Drug–Indication Ranking | Relative position of a program within the comparable drug×indication set. |
| Indication Compass | Directional signal indicating whether the program’s mechanism and profile align best with the primary indication or with alternative indication opportunities. |
| Combinability | Compatibility score for combination regimens in-indication: prediction of likely additive adverse-burden interactions. |
| Discontinuation probability | Estimated probability of program discontinuation over a defined horizon, based on longitudinal competitive pressure, space-level performance, and within-class comparables. |
| Effect estimate | Signed magnitude proxy for expected treatment effect (mechanistic efficacy vs adverse events). |
All metrics are computed outputs intended for cross-program comparability and institutional risk analysis. They are not clinical claims and do not substitute for trial readouts.
Method diagram
High-level schematic
Tip: open the image in a new tab for a higher-resolution view.
Framework steps
1. Drug–target interaction modeling
Aggregate proteome-scale interaction evidence (affinity estimates, docking, cavity mapping, ligand similarity) to form a mechanistic interaction profile.
2. Functional mechanistic translation
Translate interaction profiles into pathway and network perturbation estimates to model therapeutic mechanisms and off-target liabilities.
3. Disease matching and contraindication scoring
Compare predicted perturbations to disease signatures to identify restoration vs amplification of disease-driving biology.
4. Symptom and adverse-event alignment
Map mechanistic consequences to phenotype-level outcomes using ontology alignment and tissue-context constraints.
5. Integration and issuance
Combine evidence layers into issued, versioned outputs: a headline risk value, time-bounded curves, top drivers, and an immutable historical record.
Issuance discipline
- Scores are issued as timestamped snapshots (immutable once issued).
- Updates appear as new versions, preserving historical auditability.
- Outputs are designed for institutional reliance (committee memos, underwriting, governance).
Note: this page is intentionally high-level. Proprietary methods, weighting schemes, and model details are omitted.