On the predictability of postoperative complications for cancer patients: a Portuguese cohort study

AbstractPostoperative complications are still hard to predict despite the efforts towards the creation of clinical risk scores. The published scores contribute for the creation of specialized tools, but with limited predictive performance and reusability for implementation in the oncological context. This work aims to predict postoperative complications risk for cancer patients, offering two major contributions. First, to develop and evaluate a machine learning-based risk score, specific for the Portuguese population using a retrospective cohort of 847 cancer patients undergoing surgery between 2016 and 2018, for 4 outcomes of interest: (1) existence of postoperative complications, (2) severity level of complications, (3) number of days in the Intermediate Care Unit (ICU), and (4) postoperative mortality within 1 year. An additional cohort of 137 cancer patients from the same center was used for validation. Second, to improve the interpretability of the predictive models. In order to achieve these objectives, we propose an approach for the learning of risk predictors, offering new perspectives and insights into the clinical decision process. For postoperative complications the Receiver Operating Characteristic Curve (AUC) was 0.69, for complications’ severity AUC was 0.65, for the days in the ICU the mean absolute error was 1.07 days, and for 1-year postoperative mortality the AUC was 0.74, calculated on the development cohort. In this study, predictive models which could help to guide physicians at organizational and clinical decision making were developed. Additionally, a web-based decision support tool is further provided to this end.

Download Full-text

On the predictability of postoperative complications for cancer patients: a Portuguese cohort study

10.1101/2021.03.27.21254473 ◽

2021 ◽

Author(s):

Daniel Mateus Goncalves ◽

Rui Henriques ◽

Lucio Santos ◽

Rafael S Costa

Keyword(s):

Postoperative Complications ◽

Cancer Patients ◽

Predictive Models ◽

Characteristic Curve ◽

Postoperative Mortality ◽

Absolute Error ◽

Risk Scores ◽

Prediction Errors ◽

The Mean ◽

The Creation

Postoperative complications following cancer surgeries are still hard to predict despite the historical efforts towards the creation of standard clinical risk scores. The differences among score calculators, contribute for the creation of highly specialized tools, with poor reusability in foreign contexts, resulting in larger prediction errors in clinical practice. This work aims to predict postoperative complications risk for cancer patients, offering two major contributions. First, to develop and evaluate a machine learning-based risk score, specific for the Portuguese population using a retrospective cohort of 847 cancer patients undergoing surgery between 2016 and 2018, predicting 4 outcomes of interest: i) existence of postoperative complications, ii) severity level of complications, iii) number of days in the Intermediate Care Unit (ICU), and iv) postoperative mortality within 1 year. An additional cohort of 137 cancer patients was used to validate the models. Second, to support the study with relevant findings and improve the interpretability of predictive models. In order to achieve these objectives, a robust methodology for the learning of risk predictors is proposed, offering new perspectives and insights into the clinical decision process. For postoperative complication's the mean Receiver Operating Characteristic Curve (AUC) was 0.69, for complications severity mean AUC was 0.65, for the days in the ICU the Mean Absolute Error (MAE) was 1.07 days, and for one-year postoperative mortality the mean AUC was 0.74, calculated on the development cohort. In this study, risk predictive models which may help guide physicians at estimating cancer patient's risk of developing surgical complications were developed. Additionally, a web-based decision support system is further provided to this end.

Download Full-text

Use of risk scores for predicting new atrial fibrillation after ischemic stroke or transient ischemic attack—A systematic review

International Journal of Stroke ◽

10.1177/17474930211045880 ◽

2021 ◽

pp. 174749302110458

Author(s):

Amit K Kishore ◽

Mohammad J Hossain ◽

Alan Cameron ◽

Jesse Dawson ◽

Andy Vail ◽

...

Keyword(s):

Atrial Fibrillation ◽

Ischemic Stroke ◽

Transient Ischemic Attack ◽

Clinical Decision Making ◽

Characteristic Curve ◽

Clinical Decision ◽

Risk Scores ◽

Ecg Monitoring ◽

Discriminative Ability ◽

Ischemic Attack

Background Newly detected atrial fibrillation (NDAF) following an ischemic stroke or transient ischemic attack is often paroxysmal in nature. While challenging to detect, extended electrocardiographic (ECG) monitoring is often used to identify NDAF which has resource implications. Prognostic risk scores have been derived which may stratify the risk of NDAF and inform patient selection for ECG monitoring approaches after ischemic stroke/transient ischemic attack. Aim The overall aim was to identify risk scores that were derived and/or validated to predict NDAF after ischemic stroke/transient ischemic attack and evaluate their performance. Summary of review A systematic literature review was undertaken in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement, with application of the Quality Assessment of Diagnostic Accuracy-2 tool. Published studies, which derived and validated clinical risk scores in patients with ischemic stroke/transient ischemic attack, or externally validated an existing score to predict NDAF after ischemic stroke/transient ischemic attack, were considered and independently screened by two reviewers. Twenty-one studies involving 23 separate cohorts were analyzed from which 17 integer-based risk scores were identified. The overall frequency of NDAF was 9.7% (95% confidence intervals 8%–11.5%; I2 = 98%). The performance of the scores varied widely among derivation and validation cohorts (area under the receiver operating characteristic curve (AUC) 0.54–0.94); scores derived from stroke cohorts (12 scores) appeared to perform better (AUC 0.7–0.94) than those derived from non-stroke cohorts (five scores; AUC 0.53–0.79). The scores also varied considerably in their complexity, ascertainment, component variables, participant characteristics, outcome definition, and ease of application limiting their generalizability and utility. Conclusion Overall, the risk scores identified performed variably in their discriminative ability and the utility of these scores to predict NDAF in clinical practice remains uncertain. Further studies are required using larger prospective cohorts and randomized control trials to evaluate the usefulness of such scores for clinical decision making and preventative intervention.

Download Full-text

VEGF-Related Germinal Polymorphisms May Identify a Subgroup of Breast Cancer Patients with Favorable Outcome under Bevacizumab-Based Therapy—A Message from COMET, a French Unicancer Multicentric Study

Pharmaceuticals ◽

10.3390/ph13110414 ◽

2020 ◽

Vol 13 (11) ◽

pp. 414

Author(s):

Jocelyn Gal ◽

Gérard Milano ◽

Patrick Brest ◽

Nathalie Ebran ◽

Julia Gilhodes ◽

...

Keyword(s):

Breast Cancer ◽

Cancer Patients ◽

Clinical Decision Making ◽

Metastatic Breast ◽

Clinical Decision ◽

Risk Scores ◽

Nucleotide Polymorphisms ◽

Breast Cancer Patients ◽

Survival Prognosis ◽

Multicentric Study

The prospective multicenter COMET trial followed a cohort of 306 consecutive metastatic breast cancer patients receiving bevacizumab and paclitaxel as first-line chemotherapy. This study was intended to identify and validate reliable biomarkers to better predict bevacizumab treatment outcomes and allow for a more personalized use of this antiangiogenic agent. To that end, we aimed to establish risk scores for survival prognosis dichotomization based on classic clinico-pathological criteria combined or not with single nucleotide polymorphisms (SNPs). The genomic DNA of 306 patients was extracted and a panel of 13 SNPs, covering seven genes previously documented to be potentially involved in drug response, were analyzed by means of high-throughput genotyping. In receiver operating characteristic (ROC) analyses, the hazard model based on a triple-negative cancer phenotype variable, combined with specific SNPs in VEGFA (rs833061), VEGFR1 (rs9582036) and VEGFR2 (rs1870377), had the highest predictive value. The overall survival hazard ratio of patients assigned to the poor prognosis group based on this model was 3.21 (95% CI (2.33–4.42); p < 0.001). We propose that combining this pharmacogenetic approach with classical clinico-pathological characteristics could markedly improve clinical decision-making for breast cancer patients receiving bevacizumab-based therapy.

Download Full-text

Next-generation sequencing of BRCA1/2 in breast cancer patients: potential effects on clinical decision-making using rapid, high-accuracy genetic results

Annals of Surgical Treatment and Research ◽

10.4174/astr.2017.92.5.331 ◽

2017 ◽

Vol 92 (5) ◽

pp. 331 ◽

Cited By ~ 5

Author(s):

Hyung Seok Park ◽

Seo-Jin Park ◽

Jee Ye Kim ◽

Sanghwa Kim ◽

Jaegyu Ryu ◽

...

Keyword(s):

Breast Cancer ◽

Decision Making ◽

Next Generation Sequencing ◽

Cancer Patients ◽

Clinical Decision Making ◽

Clinical Decision ◽

High Accuracy ◽

Breast Cancer Patients ◽

Generation Sequencing ◽

Genetic Results

Download Full-text

Bayesian Network Models with Decision Tree Analysis for Management of Childhood Malaria in Malawi

10.21203/rs.3.rs-60209/v1 ◽

2020 ◽

Author(s):

Sanya B. Taneja ◽

Gerald P. Douglas ◽

Gregory F. Cooper ◽

Marian G. Michaels ◽

Marek J. Druzdzel ◽

...

Keyword(s):

Decision Tree ◽

Bayesian Network ◽

Clinical Features ◽

Predictive Models ◽

Clinical Decision Making ◽

Performance Metrics ◽

Expert Knowledge ◽

Clinical Decision ◽

Automated Method ◽

Childhood Malaria

Abstract Background: Malaria is a major cause of death in children under five years old in low- and middle-income countries such as Malawi. Accurate diagnosis and management of malaria can help reduce the global burden of childhood morbidity and mortality. Trained healthcare workers in rural health centers manage malaria with limited supplies of malarial diagnostic tests and drugs for treatment. A clinical decision support system that integrates predictive models to provide an accurate prediction of malaria based on clinical features could aid healthcare worker in judicious use of testing and treatment. We developed Bayesian network (BN) models to predict the probability of malaria from clinical features and an illustrative decision tree to model the decision to use or not use a malaria rapid diagnostic test (mRDT).Methods: We developed two BN models from data that were collected in a national survey of outpatient encounters of children in Malawi. The target diagnosis is taken as the result of mRDT. The first BN model was created manually with expert knowledge, and the second model was derived using an automated method followed by modifications guided by expert knowledge. The performance of the BN models was compared to other statistical models on a range of performance metrics. We developed a decision tree that integrates predictions from these predictive models with the costs of mRDT and a course of recommended treatment. Results: Compared to the logistic regression and random forest models, the BN models had similar accuracy of 64% but had higher sensitivity at the cost of lower specificity at the default threshold. Sensitivity analysis of the decision tree showed that at low (below 0.04) and high (above 0.4) probabilities of malaria in a child, the preferred decision that minimizes expected costs is not to perform mRDT.Conclusion: In resource-constrained settings, judicious use of mRDT is important. Predictive models in combination with decision analysis can provide personalized guidance on when to use mRDT in the management of childhood malaria. BN models can be efficiently derived from data to support such clinical decision making.

Download Full-text

Clinicians’ experiences of using and implementing a medical mobile phone app (QUiPP V2) designed to predict the risk of preterm birth and aid clinical decision making

BMC Medical Informatics and Decision Making ◽

10.1186/s12911-021-01681-w ◽

2021 ◽

Vol 21 (1) ◽

Author(s):

N. Carlisle ◽

H. A. Watson ◽

J. Carter ◽

K. Kuhrt ◽

P. T. Seed ◽

...

Keyword(s):

Decision Making ◽

Preterm Birth ◽

Cultural Context ◽

Clinical Decision Making ◽

Qualitative Interviews ◽

Scale Up ◽

Clinical Decision ◽

Clinical Trial Registry ◽

Support Tool ◽

The Impact

Abstract Background As the vast majority of women who present in threatened preterm labour (TPTL) will not deliver early, clinicians need to balance the risks of over-medicalising the majority of women, against the potential risk of preterm delivery for those discharged home. The QUiPP app is a free, validated app which can support clinical decision-making as it produces individualised risks of delivery within relevant timeframes. Recent evidence has highlighted that clinicians would welcome a decision-support tool that accurately predicts preterm birth. Methods Qualitative interviews were undertaken as part of the EQUIPTT study (The Evaluation of the QUiPP app for Triage and Transfer) (REC: 17/LO/1802) which aimed to evaluate the impact of the QUiPP app on management of TPTL. Individual semi-structured telephone interviews were used to explore clinicians’ (obstetricians’ and midwives’) experiences of using the QUiPP app and how it was implemented at their hospital sites. Thematic analysis was chosen to explore the meaning of the data, through a framework approach. Results Nineteen participants from 10 hospital sites in England took part. Data analysis revealed three overarching themes which were: ‘experience of using the app’, ‘how QUiPP risk changes practice’ and ‘successfully adopting QUiPP: context is everything’. With these final themes we appeared to have achieved our aim of exploring the clinicians’ experiences of using and implementing the QUiPP app. Conclusion This study explored different clinician’s experiences of implementing the app. The organizational and cultural context at different sites appeared to have a large impact on how well the QUiPP app was implemented. Future work needs to be undertaken to understand how best to embed the intervention within different settings. This will inform scale up of QUiPP app use across the UK and ensure that clinicians have access to this free, easy-to-use tool which can positively aid clinical decision making when caring for women in TPTL. Clinical trial registry and registration number ISRCTN 17846337, registered 08th January 2018, https://doi.org/10.1186/ISRCTN17846337.

Download Full-text

P1858Meta-analysis of performance of risks scores for predicting outcomes after percutaneous Mitraclip implantation

European Heart Journal ◽

10.1093/eurheartj/ehz748.0609 ◽

2019 ◽

Vol 40 (Supplement_1) ◽

Author(s):

T K M Wang ◽

M T M Wang

Keyword(s):

Confidence Interval ◽

Risk Stratification ◽

Clinical Decision Making ◽

Operative Mortality ◽

Data Extraction ◽

Meta Analysis ◽

Critical Role ◽

Clinical Decision ◽

Risk Scores ◽

Sts Score

Abstract Background The Mitraclip is the most established percutaneous mitral valve intervention indicated for severe mitral regurgitation at high or prohibitive surgical risk. Risk stratification plays a critical role in selecting the appropriate treatment modality in high risk valve disease patients but have been rarely studied in this setting. We compared the performance of risk scores at predicting mortality after Mitraclip in this meta-analysis. Methods MEDLINE, Embase and Cochrane databases from 1 January 1980 to 31 December 2018 were searched. Two authors reviewed studies which reported c-statistics of risk models to predict mortality after Mitraclip for inclusion, followed by data extraction and pooled analyses. Results Amongst 494 articles searched, 47 full-text articles were evaluated, and 4 studies totalling 879 Mitraclip cases were included for analyses. Operative mortality was reported at 3.3–7.4% in three studies, and 1-year mortality 11.2%-13.5% in two studies. C-statistics (95% confidence interval) for operative mortality were EuroSCORE 0.66 (0.57–0.75), EuroSCORE II 0.72 (0.60–0.85) and STS Score 0.64 (0.56–0.73). Respective Peto's odds ratios (95% confidence interval) to assess calibration were EuroSCORE 0.21 (0.14–0.31), EuroSCORE II 0.43 (0.24–0.76) and STS Score 0.36 (0.21–0.61). C-statistics (95% confidence interval) for 1-year mortality were EuroSCORE II 0.64 (0.57–0.70) and STS Score (0.58–0.64). Conclusion All scores over-estimated operative mortality, and EuroSCORE II had the best moderate discrimination while the other two scores were only modestly prognostic. Development of Mitraclip-specific scores may improve accuracy of risk stratification for this procedure to guide clinical decision-making.

Download Full-text