NextStat for Biologists & Pharma

Clinical trials, PK/PD, and treatment effects

You design clinical trials, model drug kinetics, and quantify treatment effects. NextStat gives you a fast, reproducible inference engine with survival analysis (Cox PH, Weibull, AFT), pharmacometrics (PK/PD, NLME), GLMs, and Bayesian NUTS — plus validation artifacts designed for GxP and 21 CFR Part 11 audit trails.

What You Already Know → What NextStat Calls It

Quickstart: Survival Analysis

import nextstat
import numpy as np

# Time-to-event data with right censoring
times  = np.array([4.1, 7.3, 2.0, 11.5, 6.8, 1.2, 9.4, 3.7])
events = np.array([True, True, False, True, True, True, False, True])
X = np.array([
    [1, 65], [0, 72], [1, 58], [0, 61],
    [1, 70], [0, 55], [1, 68], [0, 63],
], dtype=float)  # treatment + age

# Cox PH with Efron ties
model = nextstat.CoxPhModel(times, events, X, ties="efron")
result = nextstat.fit(model)
print(f"Hazard ratios: {np.exp(result.params)}")
print(f"Standard errors: {result.std_errors}")

# Weibull parametric survival
weibull = nextstat.WeibullModel(times, events, X)
wb_fit = nextstat.fit(weibull)
print(f"Shape (k): {wb_fit.params[0]:.3f}")
print(f"Covariate effects: {wb_fit.params[2:]}")

Quickstart: Population PK (NLME)

import nextstat

# Population pharmacokinetic model
# 1-compartment oral with log-normal random effects
subjects = [
    {"id": 1, "times": [0.5, 1, 2, 4, 8], "conc": [2.1, 5.3, 4.8, 3.1, 1.2],
     "dose": 100, "weight": 70},
    {"id": 2, "times": [0.5, 1, 2, 4, 8], "conc": [1.8, 4.7, 4.2, 2.8, 0.9],
     "dose": 100, "weight": 85},
    # ... more subjects
]

# Fit population model: fixed effects (θ) + random effects (η)
pop_fit = nextstat.nlme(
    model="pk_1cmt_oral",
    data=subjects,
    fixed={"ka": 1.0, "ke": 0.2, "V": 10.0},  # initial estimates
    random=["ka", "V"],                          # random on absorption + volume
)
print(f"Population ka: {pop_fit.fixed['ka']:.3f}")
print(f"Population V:  {pop_fit.fixed['V']:.2f}")
print(f"Omega (variance of random effects): {pop_fit.omega}")

Validation: GxP-Ready Artifacts

NextStat's validation report is designed for regulated environments. The CLI produces a deterministic JSON artifact + optional PDF suitable for:

• IQ/OQ/PQ validation packs (FDA 21 CFR Part 11)
• EMA GxP / CSA (Computer Software Assurance) evidence
• Per-suite pass/fail with dataset fingerprints (SHA-256)
• Environment lock: OS, Rust version, cargo.lock hash, Python version
• Deterministic mode: stable JSON key ordering, reproducible PDF metadata

# Generate validation report (CLI)
nextstat validation-report \
  --apex2 tmp/apex2_master_report.json \
  --workspace workspace.json \
  --out validation_report.json \
  --pdf validation_report.pdf \
  --deterministic

# Or via Makefile (local + CI)
make validation-pack

When to Use NextStat vs NONMEM / R

Next Steps

• Survival Analysis — Cox PH, Weibull, log-normal AFT → Survival Analysis
• Bayesian Sampling — NUTS/HMC with full diagnostics → Bayesian Sampling
• Regression & GLM — Logistic, Poisson, negative binomial → Regression & GLM
• Validation Report — GxP-ready JSON+PDF artifacts → Validation Report
• Agentic Tools — LLM tool definitions for AI-driven analysis → Agentic Tools
• Server API — Self-hosted GPU inference for shared compute → Server API
• Glossary — HEP ↔ DS ↔ Quant ↔ Bio term mapping → Glossary