v0.7

v0.7 – SD validation refinement

• Introduced structured covariance-aware fit diagnostics, including covariance matrix, correlation matrix, and coefficient-level 1σ estimates derived from the fit Jacobian.
• Replaced the previous fixed 1.96 conversion from 95% confidence bounds with statistically consistent coefficient uncertainty handling based on fit output and finite-DFE correction when needed.
• Added covariance-aware propagation for derived α/β uncertainty, including delta-method 95% confidence interval estimation.
• Improved fit session persistence by storing GOF and covariance-related output metadata in saved .mat sessions.
• Added backward-compatible handling of legacy sessions lacking covariance/GOF metadata, with explicit user notification when coefficient uncertainty display cannot be regenerated.
• Extended fit export reports with covariance diagnostics and optional covariance/correlation matrix output for reproducibility and downstream analysis.

v0.6

v0.6 – SD validation refinement

• Allowed non-negative SD values in data parsing routine.

v0.5

v0.5 – Robust SD handling and improved MultiSession stability

• Enforced rigorous handling of standard deviation (SD) input in the data table, ensuring consistent behavior when SD is fully absent or fully provided.
• Introduced explicit validation of partially filled or non-numeric SD columns to prevent silent propagation of invalid weights.
• Improved MultiSession viewer stability by eliminating out-of-range numeric spinner initialization during session restore.
• Minor internal clean-up to improve defensive programming and runtime stability in session import workflows.

v0.4

v0.4 – Refactor core workflow, extend RBE analysis, improve usability

• Refactored the internal application structure to improve modularity, robustness, and long-term maintainability.
• Strengthened separation between GUI logic and analytical routines, improving reproducibility and clarity.
• Extended RBE analysis to support bidirectional evaluation (fixed survival fraction and fixed dose).
• Added automatic computation of α/β ratios with consistent uncertainty propagation and confidence intervals.
• Introduced a normalized goodness-of-fit indicator (Adj SSE), equivalent to χ²/Ndof when inverse-variance weighting is used.
• Improved model-scanning output by reporting additional fit-quality metrics for all successfully converged models.
• Clarified and refined GUI controls, labels, icons, and tooltips to improve discoverability and ease of use.
• Revised multi-session viewer layout and interaction, including clearer table semantics and session-level removal.
• Improved reference selection and execution workflow for RBE computations, with explicit user control.
• Updated installation notes, platform limitations, and developer instructions following reviewer feedback.
• Updated documentation and example datasets to improve clarity and consistency with the revised workflow.

v0.3

v0.3 – Bug fixes

• Fixed unintended removal of data rows with stdDev = 0 during preprocessing.
• Prevented valid points from being discarded by MATLAB prepareCurveData due to infinite weights.
• Improved robustness of standard deviation parsing, avoiding column misalignment and NaN propagation.
• Stabilised the weight computation pipeline while preserving the semantic definition w = 1/σ².
• Ensured consistent behaviour between file-based input and GUI DataTable workflows.

v0.2

v0.2 – Feature update

• Added spinner controls in the Viewer tab for font size and scaling adjustment.

v0.1

v0.1 – Initial release

• First public release of SDAnext.
• Implemented core workflow for survival data import, validation, and preprocessing.
• Added basic visualization and fitting infrastructure for clonogenic survival analysis.