DoseSelect AI

AI-powered clinical trial dose optimization for Phase II studies, extracting dosage evidence from trial text and identifying dose options that best balance safety and efficacy.

The Problem

“Optimize Phase II clinical trial dose selection by combining evidence extraction, exposure-response modeling, and physician-guided recommendation workflows”

Organizations face these key challenges:

Dose evidence is buried in unstructured clinical documents and tables

Safety, efficacy, PK, and exposure-response data are fragmented across studies

Manual dose selection workflows are slow and difficult to reproduce

Clinical teams need transparent recommendations that preserve physician judgment

Sparse datasets in rare cancers limit conventional model robustness

Balancing efficacy, toxicity, and treatment burden is a multi-objective optimization problem

Regulated environments require explainability, provenance, and validation

Impact When Solved

Reduces manual review time for dose evidence extraction from protocols, publications, and study reportsImproves consistency of benefit-risk assessment across studies and indicationsSupports faster Phase II go/no-go and dose-ranging decisionsEnables physician-in-the-loop adoption for oncology dosing workflowsImproves traceability and auditability of dose recommendation rationaleSupports rare disease and small-cohort settings where data is sparse

The Shift

Before AI~85% Manual

Human Does

•Review Phase II reports, publications, protocols, and tables to identify dose-response evidence
•Manually extract dose levels, cohorts, safety events, and efficacy endpoints into comparison sheets
•Compare benefit-risk patterns across studies and discuss candidate doses in review meetings
•Document rationale, citations, and final dose recommendations for Phase II decisions

Automation

•No material AI support in the legacy workflow

With AI~75% Automated

Human Does

•Validate extracted dose, cohort, safety, and efficacy evidence flagged for review
•Set clinical decision criteria and review ranked dose options against program context
•Resolve ambiguous findings, data gaps, and cross-study inconsistencies requiring judgment

AI Handles

•Ingest trial reports, publications, and protocols and extract structured dose-related evidence
•Normalize dose, endpoint, and cohort data across studies and link outputs to source text
•Analyze safety-efficacy patterns and rank candidate dose options with rationale and citations
•Monitor new evidence and update recommendation summaries and auditable decision trails

Operating Intelligence

How DoseSelect AI runs once it is live

AI runs the first three steps autonomously.

Humans own every decision.

The system gets smarter each cycle.

Confidence95%

ArchetypeRecommend & Decide

Shape6-step converge

Human gates1

Autonomy

67%AI controls 4 of 6 steps

Who is in control at each step

Each column marks the operating owner for that step. AI-led actions sit above the divider, human decisions and feedback loops sit below it.

Loop shapeconverge

Step 1

Assemble Context

Step 2

Analyze

Step 3

Recommend

Step 4

Human Decision

Step 5

Execute

Step 6

Feedback

AI lead

Autonomous execution

1AI

2AI

3AI

5AI

gate

Human lead

Approval, override, feedback

4Human

6↺ Loop

AI-led step

Human-controlled step

Feedback loop

TL;DR

AI handles assembly, analysis, and execution. The human gate sits at the decision point. Every cycle refines future recommendations.

The Loop

6 steps

1AI

Assemble Context

Combine the relevant records, signals, and constraints.

instant

2AI

Analyze

Evaluate options, risk, and likely outcomes.

instant

3AI

Recommend

Present a ranked recommendation with supporting rationale.

instant

4Human checkpoint

Human Decision

A human accepts, edits, or rejects the recommendation.

hours to days

Authority gates · 1

The system must not finalize a Phase II dose recommendation without physician or clinical study team approval. [S1][S2][S3]

Why this step is human

The decision carries real-world consequences that require professional judgment and accountability.

5AI

Execute

Carry out the approved action in the operating workflow.

instant

6Feedback

Feedback

Outcome data improves future recommendations.

continuous

1 operating angles mapped

Operational Depth

Technologies

Technologies commonly used in DoseSelect AI implementations:

CURATE.AI recommendation interface/processOther

4 mentions

On-time recommendation delivery processOther

4 mentions

Per-cycle chemotherapy dosing workflowOther

4 mentions

Physician boundary-setting processOther

3 mentions

Covariate impact testingOther

2 mentions

+1 more technologies(sign up to see all)

Key Players

Companies actively working on DoseSelect AI solutions:

manual oncologist dose adjustment workflows general oncology CDSS platforms

Real-World Use Cases

Exposure-response modeling for tarlatamab dose selection in advanced SCLC

Researchers used patient drug-level data and outcomes to figure out which tarlatamab dose gives the best chance of helping small cell lung cancer patients while keeping risks acceptable.

predictive risk-outcome modelingdeployed analytical workflow within clinical development; evidence comes from a conference abstract using real phase 1/2 trial data.

10.0

AI-guided personalized Ibrutinib dose selection for Waldenström macroglobulinemia

An AI tool learned from one WM patient’s own treatment history and blood test results to suggest how much Ibrutinib he should take over time, instead of relying only on a one-size-fits-all dose.

closed-loop dose-response optimizationearly clinical feasibility / single-patient case study within a trial

10.0

Human-in-the-loop AI dosing workflow for palliative solid tumor care

Doctors do not hand control to the AI; they choose what data the system uses and set safe operating limits, while the AI helps suggest doses during treatment.

decision support with human oversightreal clinical workflow tested prospectively, but still early and limited to feasibility evidence.

10.0