The bioanalytical phase is the part of a BA/BE study where the molecule's concentration in plasma becomes a number on a page. Everything upstream — protocol, ethics, dosing, sample collection — produces the samples. Everything downstream — PK calculation, statistics, reporting — consumes the numbers. The bioanalytical phase is the conversion. And the conversion is where the discipline either holds or doesn't.
The architecture inside the bioanalytical stage breaks into three operational streams plus a layered QA system that audits all three. Reading them in order — what happens, then who's watching:
/ 01The operational streams.
Three streams run end-to-end inside the bioanalytical phase. They overlap in time but each has its own deliverables, its own SOPs, its own raw-data trail.
Method validation.
- Method development — selectivity, sensitivity, range, prototype runs
- Pre-validation — fit-for-purpose check before formal validation begins
- Full validation — selectivity, accuracy, precision, linearity, matrix effect, stability, dilution integrity
- Validation report — the document the regulator inspects first
Study sample analysis.
- Sample receipt — chain of custody, condition check, storage
- Production runs — calibration curve, QC samples, study samples, run acceptance
- Incurred sample reanalysis (ISR) — 10% reanalysis in pivotal studies
- Failed-run disposition — documented reanalysis authority and rationale
Data processing.
- Data extraction & review — from instrument software to processed concentration tables
- Data reporting — bioanalytical report drafted with all method, validation, run, and ISR data
- Quality control review — internal QC scientist signs off on the report
- QA review — quality assurance audits the full bioanalytical package
- Sponsor review — concentration data and report dispatched to sponsor
- Data transfer to PK/Stat analysis — the formal handover to the next stage
/ 02The QA layer that watches all three.
Quality assurance for bioanalytical operations runs in two modes — real-time (system check & in-process review) and retrospective (post-run review of raw data and reports). Both modes run continuously across all three streams above. The QA architecture is what the regulator audits when an inspection visits.
System check · real-time.
- In-process audit for trial conduct — observation during run
- System compliance verification — SOPs, deviations, training
- IT-QA · compliance verification of computerised systems
- CSV · software compliance verification (chromatography software, LIMS, EDC)
- Calibration & preventive maintenance verification — instrument fitness
- Training-compliance verification — analyst qualification matrix
- Spreadsheet use & compliance verification — the often-skipped audit
Document · raw data review.
- Method SOP & protocol review — before validation begins
- Method validation raw data review — chromatograms, integration, reweighing
- Method validation report review — report against raw data
- Study analysis raw data review — per-run audit of accepted runs
- Pre-stat concentration data review — before handover to PK
- Study analysis report review — final BA report against raw data
- DBE table / Annexure 2.7.1 review — sponsor-facing document
/ 03What the flow produces.
The bioanalytical flow has one downstream customer: the statistical and PK analysis stream that converts concentrations into Cmax, AUC, and the geometric mean ratio against the reference. If the concentration data is clean — meaning calibration curve passed every run, QCs passed every run, ISR passed, deviations dispositioned — the PK/stat analysis is mechanical. If the concentration data has any unexplained gaps, the PK analysis becomes an investigation, and the investigation becomes the reason the study is rejected.
Bioanalytical is where most regulatory observations are written because it is where the most decisions get made.This is also why the bioanalytical phase is where AI-augmented operations — peak-integration assistance, anomaly flagging, automated chromatogram review — are arriving fastest. Each of the QA touchpoints above is a candidate for AI assistance, and each of them already has a regulator-facing inspection trail that the AI must satisfy. The vaccine model from the manifesto applies here directly: AI deployed inside the bioanalytical flow must be vaccinated before deployment, monitored during, and retrained from each incident.