Clinical review
Manual subject, demographics, randomisation, inclusion/exclusion and protocol-record review in both editions.
Detect potential data anomalies, independently verify PK and ANOVA results, and resolve data-integrity concerns before regulatory review.
BEADS is designed to support authority-specific study designs, acceptance criteria, and guided study setup in accordance with regulatory frameworks used by agencies across the world.
Four connected review layers help sponsor teams move from study setup to evidence-based acceptance, query or escalation.
Manual subject, demographics, randomisation, inclusion/exclusion and protocol-record review in both editions.
Manual review in Standard. Professional adds automated source-data verification and evidence reconciliation.
NCA, descriptive statistics, ANOVA, GMR, confidence intervals and BE decisions are included in both editions.
Run the common 25-test PK/BE integrity suite and document each flag, finding, limitation and follow-up action.
Both plans include the same PK, BE statistics and 25-test integrity core. Professional is recommended for sponsors requiring automated bioanalytical verification and evidence reconciliation.
For sponsor teams that will perform clinical and bioanalytical review manually.
For sponsor teams that want automated bioanalytical verification while retaining manual clinical review.
| Annual study pack | Standard | Professional · Recommended | Sponsor-side access |
|---|---|---|---|
| 5 studies/year | $2,500First year: $2,000 |
$3,000First year: $2,400 |
1 Sponsor Admin
+
1 Additional User
|
| 10 studies/year | $4,500First year: $3,600 |
$5,500First year: $4,400 |
1 Sponsor Admin
+
2 Additional Users
|
| 15 studies/year | $6,000First year: $4,800 |
$7,500First year: $6,000 |
1 Sponsor Admin
+
3 Additional Users
|
| 20 studies/year | $7,000First year: $5,600 |
$9,000First year: $7,200 |
1 Sponsor Admin
+
4 Additional Users
|
| 25 studies/year | $7,500First year: $6,000 |
$10,000First year: $8,000 |
1 Sponsor Admin
+
5 Additional Users
|
Use the three plain-language filters below. The Professional column is highlighted because it is the recommended sponsor plan.
The three platforms serve different primary purposes. BEADS is designed as a guided sponsor-side BE review and data-integrity workflow; SAS is a broad, highly programmable statistical environment; and Phoenix WinNonlin is a specialist PK/PD, NCA and bioequivalence analysis platform.
| Feature | BEADS | SAS | Phoenix WinNonlin |
|---|---|---|---|
| Primary purpose | Integrated sponsor review of clinical, bioanalytical, PK/BE and data-integrity evidence before submission. | General-purpose statistical programming, data management, modelling and reporting across many industries and study types. | Specialist pharmacokinetic, pharmacodynamic, toxicokinetic, NCA and bioequivalence analysis. |
| User workflow | Guided, study-based workflow with agency setup, structured review steps, findings and sponsor reports. | Analyses are created through SAS procedures, code, macros and organisation-specific programs. | Graphical project workflow using worksheets, object mappings, NCA, modelling and bioequivalence objects. |
| Programming | No coding for the configured BE review workflow. | Programming is normally required to prepare data, specify models, automate outputs and maintain validated programs. | No coding is required for standard graphical workflows, although correct object configuration and mappings are required. |
| Expertise required | Guided for sponsor teams; qualified scientific review remains necessary for findings and conclusions. | Requires statistical-programming and model-selection expertise. | Requires PK/NCA and bioequivalence expertise to configure analyses and interpret outputs correctly. |
| Data import and mapping | Excel and CSV import with guided mapping, validation and study-specific organisation. | Supports broad data access and import, but preparation, restructuring and quality checks are generally programmed. | Imports structured worksheets/files and requires column mapping to the selected analysis object. |
| Data organisation | Automatically organises mapped data into the sponsor study workflow and review modules. | Fully flexible, but organisation and reconciliation depend on the user’s programs and standards. | Project- and worksheet-based organisation with analysis objects and mapped variables. |
| Study designs | Guided support for standard crossover, replicate and parallel BE designs, with configured regulatory pathways. | Can analyse major and customised designs when the appropriate statistical model is programmed. | Average BE workflows support nonreplicated crossover, replicated crossover and parallel designs; additional individual/population BE workflows are also available. |
| PK / NCA analysis | Integrated PK calculation and sponsor review within the same BE investigation. | Possible through programmed calculations, procedures or validated organisation-specific workflows. | Core strength: dedicated NCA and PK/PD/TK analysis with integrated data processing, tables and graphics. |
| ANOVA and confidence intervals | Automated, design-aware BE analysis with guided GMR, confidence-interval and conclusion presentation. | Powerful and highly customisable ANOVA, GLM and mixed-model capability when correctly programmed. | Integrated bioequivalence models, confidence intervals and model outputs based on the selected study design and mappings. |
| Power and sample-size assessment | Included in the guided BE workflow for configured designs. | Available through SAS power and sample-size procedures, including equivalence analyses, with user-defined inputs and code. | Provides TOST power output and supports planning workflows; setup and interpretation remain user-controlled. |
| Clinical review | Dedicated sponsor workspace for subject, demographic, randomisation, inclusion/exclusion and related review. | No dedicated out-of-the-box sponsor BE clinical-review workspace; can be built through custom data and reporting programs. | Not primarily designed as a clinical source-data review platform. |
| Bioanalytical review | Manual review in Standard and automated bioanalytical verification in Professional, including MV and study-result consistency checks. | Can analyse bioanalytical data when suitable datasets, rules and programs are created, but the workflow is not purpose-built by default. | Concentration data can feed PK analyses, but full method-validation and study-source integrity review is not its primary purpose. |
| Data-integrity and anomaly screening | Integrated PK integrity scorecard plus method-validation and bioanalytical study-result checks, with affected records and investigation outputs. | Custom anomaly tests can be programmed, but the user must design, validate, maintain and interpret the complete framework. | Provides analytical and diagnostic outputs, but is not positioned as an integrated sponsor-side BE data-integrity investigation suite. |
| Sponsor collaboration and follow-up | Findings, reviewer roles, CRO questions, evidence requests and sponsor conclusions are managed in one platform. | Requires separate workflow, document-management or reporting systems unless custom-developed. | Primarily an analysis environment; sponsor queries and investigation follow-up normally occur outside the application. |
| Reports and traceability | Sponsor-focused reports, findings, review status, study credits, user roles and controlled workflow history. | Highly configurable reporting and reproducible code logs, subject to the organisation’s validation and governance controls. | Analysis outputs, worksheets, tables, charts and project history within the Phoenix project environment. |
| Best suited for | Sponsors seeking one guided environment to review BE evidence, independently verify results and investigate integrity signals. | Organisations needing maximum statistical flexibility and having experienced programmers and validated coding standards. | Scientists needing an established specialist environment for PK/NCA, modelling and conventional BE analysis. |
Capabilities can vary by product version, licence and organisational configuration. BEADS does not replace qualified scientific review, and third-party product capabilities should be confirmed before public commercial use.
Each area is presented separately so sponsors can understand the purpose, review logic and significance of the available checks. Findings are investigation signals and do not independently prove data manipulation.
| # | Test name | Classification | How it supports anomaly detection |
|---|---|---|---|
| 1 | Boxplot / Interquartile Range (IQR) | Outlier Analysis | Flags PK values (e.g., Cmax, AUC) that fall beyond 1.5× the interquartile range. Fabricated or “cleaned” values often sit suspiciously just inside or outside this fence — close enough to look normal, but not quite genuine. |
| 2 | Z-Score (Standardised Residual) | Outlier Analysis | Measures how many standard deviations a value is from the group mean. A large Z-score (>2 or >3) means the value doesn’t fit the normal spread of the rest of the subjects. |
| 3 | Robust Z-Score (Median / MAD) | Outlier Analysis | A tougher version of the Z-score that uses the median instead of the mean, so it isn’t distorted by other outliers already present. Still catches manipulation even when several outliers exist in the same dataset. |
| 4 | Grubbs’ Test (ESD) | Outlier Analysis | A formal statistical test that proves whether the single most extreme point truly does not belong to the dataset — useful when a value appears to have been altered to help a study pass or fail. |
| 5 | Dixon’s Q Test | Outlier Analysis | Compares a suspicious value against its nearest neighbours. Sensitive enough to catch even one altered value in small BE datasets (24–36 subjects), where manipulation is easiest to hide. |
| 6 | Studentized Residual | Outlier Analysis | Standardises the leftover error after fitting a regression line. A value manually adjusted to “fit” a trend shows an abnormally large residual once properly standardised — exposing forced-fit data. |
| 7 | β Slope (Linear Regression) | Similarity Check | Fits a straight-line trend through paired data and checks the slope. Unnatural slope deviations appear when real biological variability has been replaced with “cleaner”, smoothed numbers. |
| 8 | β Slope (Scaling Pattern) | Similarity Check | Checks whether one subject’s values are a near-perfect scaled version of another’s. Perfect or near-perfect linear scaling is a red flag, since real biological data is never that clean. |
| 9 | Average Percentage Difference (APD) | Similarity Check | Calculates the average percentage difference between paired values across subjects or arms. An unusually small or consistent APD suggests values were copy-adjusted rather than independently measured. |
| 10 | f₂ Similarity Factor | Similarity Check | A standard measure of how similar two profiles are. A suspiciously high f₂ (too similar) suggests one profile was derived or copied from another rather than independently measured. |
| 11 | Wilcoxon Rank Sum | Non-parametric Test | Compares the ranking (order) of two groups of values rather than their raw numbers. Manipulated data tends to cluster in rank in ways that don’t reflect true random biological sampling. |
| 12 | Tmax Clustering Detection | Non-parametric Test | Looks at how many subjects share the exact same Tmax (time of peak concentration). Excessive identical Tmax values across “different” subjects suggest numbers were copied or interpolated rather than observed. |
| 13 | EDA — ln(Cmax) Bar Analysis | Trend Analysis | Visually and statistically compares log-transformed Cmax values across subjects. Bars that are too uniform, or that repeat a pattern, indicate templated or copied data. |
| 14 | Runs Test (Non-Parametric) | Trend Analysis | Checks whether increases and decreases in a sequence occur in a truly random order. Detects non-random patterns — e.g., every alternate subject being “adjusted” — that shouldn’t exist in real data. |
| 15 | Sequential CI and GMR Analysis | Trend Analysis | Tracks how the confidence interval and GMR evolve as subjects are added one by one. An unnatural “jump” to pass criteria near the final subjects, instead of gradual convergence, signals manipulation. |
| 16 | PCA — Multidimensional Detection | Trend Analysis | Reduces many PK variables into a few dimensions to visualise how subjects cluster together. Manipulated subjects can show up as a separate cluster even when no single variable looks abnormal alone. |
| 17 | K-Means Clustering | Trend Analysis | Groups subjects automatically based on overall similarity across parameters. Manipulated subjects often form an unnaturally tight or separate cluster instead of blending into organic, overlapping groups. |
| 18 | Group-wise Comparative Analysis | Trend Analysis | Compares statistics across sequence, period, or site sub-groups. A mismatch in group-level statistics suggests manipulation was targeted at a specific subgroup. |
| 19 | ANOVA — ln(T/R) Variance | Trend Analysis | Examines the residual (unexplained) variance from the standard BE ANOVA model. Artificially low, unexplained variance suggests real biological noise was suppressed. |
| 20 | Residual Analysis | Trend Analysis | Studies the pattern left over after fitting the expected statistical model. Fabricated points often leave a non-random pattern in residuals, unlike genuine measurement error. |
| 21 | ISCV Consistency Check | Trend Analysis | Checks whether the intra-subject coefficient of variation (ISCV) is realistic for the drug and analytical method. An implausibly low ISCV suggests smoothing or fabrication, since real variability rarely comes out that tight. |
| 22 | Terminal Phase Quality (λz R²) | Trend Analysis | Checks the goodness-of-fit (R²) of the terminal elimination phase used to calculate λz. A suspiciously perfect R² (0.999+) across many subjects suggests values were smoothed or back-calculated. |
| 23 | Benford’s Law — First Digit | Trend Analysis | Compares the distribution of leading digits in the dataset to the expected natural pattern. Fabricated numbers rarely follow Benford’s expected leading-digit distribution. |
| 24 | Bootstrapped CI Stability | Trend Analysis | Resamples the dataset repeatedly to see how stable the confidence interval remains. An unstable CI on resampling suggests a small number of manipulated points are driving the result. |
| 25 | Period Outlier Asymmetry | Trend Analysis | Checks whether outliers are evenly spread across study periods or concentrated in just one. Outliers clustering heavily in one period only suggest selective adjustment tied to formulation or period. |
| # | Review test / area | What BEADS checks | Why it matters |
|---|---|---|---|
| 1 | Protocol-to-report coverage | Compares planned validation experiments and acceptance criteria with the final MV report. | Helps identify omitted experiments, incomplete reporting, or unsupported conclusions. |
| 2 | Calibration range, model and weighting | Reviews the validated range, curve model, weighting, accepted points, and actual study use. | Confirms that the study remains within the validated analytical conditions. |
| 3 | Accuracy and precision | Reconciles within-run and between-run performance with experiment-level results. | Identifies inconsistencies between source results and reported summary statistics. |
| 4 | Selectivity and sensitivity | Reviews blank, zero, LLOQ, interference, and sensitivity evidence. | Supports confidence that the method can distinguish and quantify the analyte reliably. |
| 5 | Recovery and matrix effect | Checks recovery and matrix-effect results across lots, levels, and experimental conditions. | Highlights selective reporting or unsupported generalisation across matrices. |
| 6 | Stability verification | Reconciles bench-top, freeze-thaw, processed-sample, stock, and long-term stability results. | Confirms that reported sample handling and storage are supported by validation evidence. |
| 7 | Repeat and selective-reporting review | Identifies unexplained repeats, excluded experiments, or inclusion of only favourable outcomes. | Supports investigation of incomplete or outcome-directed reporting. |
| 8 | MV-to-study applicability | Compares the validated method with actual study use, laboratory, matrix, equipment, and sample preparation. | Identifies possible partial-validation, cross-validation, or unsupported-use triggers. |
| # | Review test / area | What BEADS checks | Why it matters |
|---|---|---|---|
| 1 | Sequence and acquisition chronology | Reviews injection order, timestamps, missing or duplicate injections, sequence gaps, and calibration/QC placement. | Identifies chronology concerns and incomplete electronic-history evidence. |
| 2 | Run composition and acceptance | Reconciles each analytical run, accepted or rejected status, QC distribution, and bracketing. | Checks whether the reported run decision is supported by the complete run inventory. |
| 3 | Calibration and QC reconciliation | Compares source calibration and QC values, exclusions, calculations, and final run conclusions. | Highlights unsupported exclusions, altered summaries, or incorrect acceptance decisions. |
| 4 | Internal-standard and retention-time behaviour | Reviews IS response, retention-time drift, assignment patterns, and repeated chromatographic behaviour. | Helps detect unusual analytical patterns requiring chromatogram or audit-trail review. |
| 5 | Carryover assessment | Checks high-to-blank positions, predose findings, and carryover-decay patterns. | Supports evaluation of whether reported concentrations may be influenced by preceding injections. |
| 6 | Duplicate or reused data screening | Screens areas, ratios, concentrations, retention times, and numerical or file fingerprints for unexplained similarity. | Identifies record pairs that require source-file and preparation-record investigation. |
| 7 | Concentration-result reconciliation | Compares source results, report tables, transfer files, and the final PK input record by record. | Detects omitted records, mapping errors, or downstream changes in concentration data. |
| 8 | Repeat, reanalysis, reinjection and ISR | Reviews the original-to-final chronology, reasons, decision rules, and selected final values. | Helps identify unexplained repeats, directional result selection, or inconsistent ISR handling. |
Use the arrows or horizontal scroll. Select any image to open it at full size without leaving the website.
Study status, active investigations, and analysis modules in one sponsor workspace.
Set product, CRO/site, sponsor code, and the review modules required for the study.
Choose the submission environment and carry the applicable study-design and BE pathway through the workflow.
Configure crossover, replicate, parallel, HVD, and NTI study options.
Review GMRs, confidence intervals, acceptance limits, and the final BE conclusion.
Review flags, confidence levels, next steps, and evidence-linked investigation findings.
Review subject data, demographics, randomisation, and inclusion/exclusion information.
Review method validation, calibration, QC, stability, ISR, and source-data evidence.
The Sponsor Admin controls the organisation workspace and can invite role-based users within the purchased package.
Organisation, role, subscription status, study credits, users, invitations, and documents at a glance.
No sponsor can see another sponsor's studies. BEADS personnel have no routine access to sponsor study data. Any exceptional support access must be explicitly authorised by the Sponsor Admin, limited to the approved purpose and time, and recorded in the audit history.
Organisation-level tenant separation prevents cross-sponsor visibility of studies, users, findings, documents and reports.
The Sponsor Admin controls invitations and assigns Analyst, Reviewer, Study Manager or Read Only access according to responsibilities.
Support staff do not browse sponsor data. Exceptional access requires sponsor approval, a defined support reason and logged activity.
Production deployment is expected to use encryption in transit and at rest, secure upload channels and controlled download/report permissions.
User actions, study usage, findings, reviews, report versions and authorised access events are retained for traceability.
Retention, backup, archive and secure deletion periods are governed by the subscription agreement and sponsor-approved data-processing terms.
Clear answers for sponsors evaluating BEADS and preparing for a trial or subscription discussion.
Both editions include the same PK analysis, BE statistics, ANOVA and 25-test integrity scorecard. Standard uses manual clinical and manual bioanalytical review. Professional keeps clinical review manual and adds automated bioanalytical source-data verification and evidence reconciliation.
The launch offer reduces the selected annual subscription price by 20% for the first subscription year. The standard annual price is displayed alongside the discounted first-year amount.
Every package includes one Sponsor Admin. The 5-study package adds 1 user; 10 studies adds 2 users; 15 studies adds 3 users; 20 studies adds 4 users; and 25 studies adds 5 users.
The Sponsor Admin can invite additional users and assign them as Analyst, Reviewer, Study Manager or Read Only, subject to the number of users included in the selected package.
No. A flag identifies an unusual value, pattern, inconsistency or source-data concern that requires scientific and documentary investigation. The sponsor should review the affected records and obtain appropriate evidence or explanation from the CRO.
BEADS is designed to review clinical information, bioanalytical method-validation evidence, bioanalytical study results, concentration-time data, PK calculations, ANOVA and bioequivalence conclusions. The exact assessment depends on the source files supplied.
The platform is designed around separate sponsor workspaces and role-based access. Users from one sponsor organisation should not be able to access another sponsor’s studies. Final security commitments will be governed by the production architecture and subscription/data-processing agreements.
The BEADS team will review the submitted details and contact the sponsor to understand the expected study volume, review requirements and preferred edition. As the platform is in final testing, the initial interaction may include a guided preview before trial access is activated.
No. BEADS supports structured review, independent calculations and anomaly identification. Final interpretation, CRO queries and submission decisions remain the responsibility of qualified sponsor personnel.
BEADS is in the final stage of testing. The fully functional and tested software will be launched after completion of verification. Meanwhile, interested sponsors are invited to review the current interface and share their preferred edition and annual study pack.
Embedded demonstration placeholder. The narration script below is designed for a video of approximately three minutes.
Illustrative embedded demo. Replace this source with the final narrated BEADS video.
Bioequivalence studies generate clinical, bioanalytical, pharmacokinetic and statistical evidence that a sponsor must trust before submission. BEADS was developed to help sponsors investigate that evidence before a regulatory reviewer does. The platform screens for potential data anomalies, independently reproduces PK and ANOVA results, and brings the review into one controlled workflow.
A sponsor begins by creating an investigation, selecting the applicable regulatory framework and matching the protocol to the correct study design. BEADS supports the major crossover, replicate and parallel bioequivalence designs, including high-variability and narrow-therapeutic-index pathways where applicable.
Clinical data can be reviewed through subject, demographic, randomisation and eligibility workspaces. The bioanalytical module covers method validation, calibration curves, quality controls, stability, ISR and study-result evidence. In the Professional edition, automated bioanalytical verification helps reconcile supplied source data with the reported conclusion.
For PK and bioequivalence, BEADS calculates key PK parameters, descriptive statistics, ANOVA, geometric least-squares means, confidence intervals, power and the final BE conclusion. Both Standard and Professional editions use the same PK, statistics and data-integrity core.
The PK Data Integrity Scorecard applies 25 statistical tests. These include IQR and Z-score outlier checks, profile-similarity measures, runs tests, sequential confidence-interval analysis, PCA, clustering, residual analysis, ISCV consistency, terminal-phase quality, Benford screening and bootstrap stability. Each signal is intended to guide investigation and does not, by itself, prove manipulation.
Regulatory tools such as FDA DABERS and the European SaToWIB and Buster routines demonstrate the importance of anomaly detection, cumulative statistical review and PK-profile similarity. BEADS is independently developed and is not affiliated with or endorsed by those agencies. Its difference is the broader sponsor workflow combining clinical review, bioanalytical evidence, PK and ANOVA reproduction, integrity screening, observations and reporting.
Subscriptions are available as Standard or Professional annual study packs. The Sponsor Admin manages the organisation account, study credits and documents, and can invite users as Analysts, Reviewers, Study Managers or Read Only users according to the purchased package.
BEADS is currently in the final stage of testing. To view the software and discuss trial access, select Request for Trial, complete the short form and submit your interest. The BEADS team will contact you to arrange the next step. Investigate early, resolve concerns with evidence, and approach regulatory submission with greater confidence.
Complete the form to register your interest in a BEADS trial. Our team will contact you shortly to discuss access, study requirements and onboarding.
Your trial request has been recorded. The BEADS team will connect with you shortly to discuss the trial and next steps.