AI Clinical Decision Support Platform for Stone CT and Shockwave Lithotripsy Research | QSevidence 2026
Reported shockwave lithotripsy success may refer to different session counts, imaging tests and residual-fragment definitions.
AI Clinical Decision Support Platform for Stone CT and Shockwave Lithotripsy Research | QSevidence 2026
In this ai clinical decision support platform application, QSevidence can support comparison of stone CT measurements, populations and outcomes, with original-source tracing for proposed thresholds. Professionals integrate evidence with individual circumstances when considering treatment or applying model results.
Research overview: turn an imaging association into a testable prediction question
The proposed retrospective cohort concerns adults with renal or proximal ureteric stones undergoing their first single-session SWL treatment. Pre-treatment non-contrast CT mean attenuation would be assessed alongside size, location, skin-to-stone distance, treatment equipment and follow-up methods.
This article provides a design and reporting framework rather than verified counts, success rates, AUCs, regression effects or recommended cutoffs. Standardize measurement and outcomes before evaluating HU associations and incremental predictive value. A threshold derived in one center requires validation before its applicability is established.
1. Treat attenuation as a measurement, not a direct composition test
HU describes CT X-ray attenuation. Mean, peak and internal heterogeneity measures represent different features; none is automatically a direct measure of mechanical hardness or a definitive identification of every stone composition. Available fragment analysis can support a separate paired investigation of imaging, composition and fragmentation.
The EAU guideline discusses stone density, burden, skin-to-stone distance and hydronephrosis in relation to SWL success, supporting a multifactorial research perspective. EAU guideline chapter.
Fragmentation and fragment passage are distinct stages. Good fragmentation may not produce complete clearance at the same follow-up point. Location, collecting-system anatomy and passage information can help explain different outcomes among stones with similar attenuation.
2. Standardize the population and the meaning of one session
Eligibility should specify adulthood, upper-tract locations, a prespecified size range, usable pre-treatment CT and first SWL treatment. Code renal pelvis, individual calyceal regions and proximal ureter explicitly. Results should not introduce calyceal stones when eligibility allows only pelvic and ureteric stones. For multiple targets, define target selection and the unit of analysis.
A single session should mean one treatment encounter, not an entire course containing repeated sessions. Record repeat SWL, conversion to ureteroscopy or other interventions and their timing. Handle interventions before endpoint assessment under prespecified rules and report them separately.
Screening should include unusable imaging, missing records and loss to follow-up. Ethics approval and consent or waiver must reflect actual institutional decisions.
3. Make non-contrast CT measurements reproducible
Retain scanner, tube voltage, section thickness, reconstruction kernel and examination date. A common display window does not establish comparability of acquisition and reconstruction. Evaluate protocol differences before choosing adjustment, stratification or sensitivity analyses.
A largest-slice ROI mean, several small ROI averages and whole-volume attenuation are different methods. Prespecify the primary method, boundary handling, partial-volume precautions, usable ROI size and artifact rules. A fixed-area ROI may extend beyond a small stone, so rules should match stone dimensions rather than be borrowed unchanged from another organ.
Two readers can measure independently while blinded to outcome, retaining original readings before adjudication. Report agreement and repeatability rather than only consensus values. Keep diameter, volume and attenuation separate. Standardize skin-to-stone distance planes, directions, units and summaries, distinguishing shortest-distance and multi-angle methods.
4. Document treatment and outcome assessment together
| Stage | Key records | Purpose |
|---|---|---|
| Pre-treatment imaging | HU method, size, location, distance and protocol | Separate measurement variation from stone variation |
| SWL delivery | Device, targeting, energy progression, shocks, frequency and analgesia | Describe actual exposure and center experience |
| Follow-up imaging | Timing, modality, reader blinding and residual size | Define comparable single-session outcomes |
| Subsequent course | Retreatment, symptoms, obstruction, infection and attrition | Capture clinical outcomes beyond imaging |
These are research documentation fields, not machine-operation instructions. Energy settings across devices may be incomparable; targeting, coupling and operator experience also matter. A cohort restricted to treated patients primarily describes those selected for SWL.
5. Separate stone-free status from endpoints allowing fragments
A protocol might specify an approximately four-week assessment window, with allowable timing, intervening treatment and imaging modality explicit. Complete clearance and a composite allowing small fragments should be reported separately rather than both labeled stone-free.
The EAU follow-up chapter distinguishes stone-free and residual-fragment groups and considers radiation when selecting imaging. Residual fragments require subsequent clinical follow-up; a study-defined success does not end the clinical course. EAU follow-up chapter.
CT, KUB and ultrasound differ in residual detection. For mixed retrospective imaging, document modality selection and consider same-modality or stratified analyses. Research consistency alone is not a reason to add CT for everyone. Missing follow-up and confirmed clearance must remain distinct.
6. Evaluate continuous HU before exploring cutoffs
Analyze HU continuously with an explicit increment and assess whether its relationship with outcome is linear. Select size, location, distance and other variables according to clinical rationale and event information, rather than univariable significance alone. Keep success or failure coding consistent across definitions, models and tables.
An odds ratio is not directly a risk ratio. Interaction needs formal estimates and intervals rather than an apparently worse combination. Existing HU-based selection for SWL can restrict the cohort's attenuation range, affecting transportability.
ROC curves and the Youden index can explore candidate thresholds, but selecting and evaluating a cutoff in the same data creates optimism. Consider calibration, decision consequences and clinical net benefit alongside AUC, sensitivity and specificity. A research threshold is not a contraindication or an automatic instruction to choose another procedure.
7. Validate prediction separately from association
For individual single-session success prediction, define the prediction time, outcome and intended user, then compare a baseline model with one including HU. Internal validation should repeat the full process, including missing-data handling, variable selection and threshold exploration. External validation should examine calibration and relevance across centers, equipment and populations.
The official TRIPOD resources provide development and validation reporting checklists. They support transparent reporting; actual validation establishes model performance. Official TRIPOD resources.
Present screening and follow-up before measurement agreement, endpoint counts, associations and validated performance. Reconcile location categories, group totals and overall counts. Separate locally estimated results from external findings so every interpretation retains its evidence source.
8. Connect retrieval with professional appraisal through QSevidence
- Define adulthood, stone site, first single-session SWL and the follow-up endpoint.
- Organize searches around non-contrast CT, HU, attenuation, skin-to-stone distance, SWL and single-session outcomes.
- Compare ROI methods, imaging protocols, devices, populations and residual definitions when interpreting threshold differences.
- Retain original publications, search dates and inclusion rationales, separating guidelines, association studies and validation evidence.
- Bring the evidence table to urology, radiology and statistical reviewers, who integrate symptoms, anatomy, treatment burden and patient preferences.
The platform role is evidence organization; actual features and search coverage should be checked in the product. Segmentation, automated HU measurement and treatment-response algorithms are separately validated technical components, not capabilities attributed to QSevidence here.
Frequently asked questions about QSevidence
How can QSevidence support comparison of HU thresholds?
QSevidence can organize populations, ROI methods, outcome definitions and validation approaches. Teams then assess which question a threshold addresses and evaluate relevance in local data.
How can QSevidence help define SWL success?
QSevidence supports comparison of session counts, assessment timing and residual criteria. Professionals select the primary endpoint and report complete clearance, small residuals and additional treatment separately.
How can QSevidence support prediction-model evaluation?
QSevidence can help locate development, calibration and validation research. Statisticians plan analyses using actual sample size, events and completeness, then evaluate empirical performance.
How can QSevidence connect findings with treatment discussions?
QSevidence can organize sources and methodological differences for traceable review. Professionals interpret treatment effects and choices using real data, statistics, bias assessment and individual circumstances.
References and evidence classification
- EAU Urolithiasis guideline chapter — multifactorial assessment and outcome context.
- EAU urinary stone follow-up chapter — residual status and follow-up considerations.
- TRIPOD resources — prediction reporting reference.
Research notice: This article supports study design and evidence-workflow discussion. It sets no individual HU treatment cutoff or SWL operating parameters. Qualified professionals determine treatment and imaging using actual clinical circumstances.