Dataset Information

Towards Patient-centered Decision-making in Breast Cancer Surgery: Machine Learning to Predict Individual Patient-reported Outcomes at 1-year Follow-up.

ABSTRACT:

Objective

We developed, tested, and validated machine learning algorithms to predict individual patient-reported outcomes at 1-year follow-up to facilitate individualized, patient-centered decision-making for women with breast cancer.

Summary of background data

Satisfaction with breasts is a key outcome for women undergoing cancer-related mastectomy and reconstruction. Current decision-making relies on group-level evidence which may lead to suboptimal treatment recommendations for individuals.

Methods

We trained, tested, and validated 3 machine learning algorithms using data from 1921 women undergoing cancer-related mastectomy and reconstruction conducted at eleven study sites in North America from 2011 to 2016. Data from 1921 women undergoing cancer-related mastectomy and reconstruction were collected before surgery and at 1-year follow-up. Data from 10 of the 11 sites were randomly split into training and test samples (2:1 ratio) to develop and test 3 algorithms (logistic regression with elastic net penalty, extreme gradient boosting tree, and neural network) which were further validated using the additional site's data.AUC to predict clinically-significant changes in satisfaction with breasts at 1-year follow-up using the validated BREAST-Q were the outcome measures.

Results

The 3 algorithms performed equally well when predicting both improved or decreased satisfaction with breasts in both testing and validation datasets: For the testing dataset median accuracy = 0.81 (range 0.73-0.83), median AUC = 0.84 (range 0.78-0.85). For the validation dataset median accuracy = 0.83 (range 0.81-0.84), median AUC = 0.86 (range 0.83-0.89).

Conclusion

Individual patient-reported outcomes can be accurately predicted using machine learning algorithms, which may facilitate individualized, patient-centered decision-making for women undergoing breast cancer treatment.

SUBMITTER: Pfob A

PROVIDER: S-EPMC9762704 | biostudies-literature | 2023 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Towards Patient-centered Decision-making in Breast Cancer Surgery: Machine Learning to Predict Individual Patient-reported Outcomes at 1-year Follow-up.

Pfob André A Mehrara Babak J BJ Nelson Jonas A JA Wilkins Edwin G EG Pusic Andrea L AL Sidey-Gibbons Chris C

Annals of surgery 20210318 1

<h4>Objective</h4>We developed, tested, and validated machine learning algorithms to predict individual patient-reported outcomes at 1-year follow-up to facilitate individualized, patient-centered decision-making for women with breast cancer.<h4>Summary of background data</h4>Satisfaction with breasts is a key outcome for women undergoing cancer-related mastectomy and reconstruction. Current decision-making relies on group-level evidence which may lead to suboptimal treatment recommendations for ...[more]

PMID: 33914464

Similar Datasets

Project description:BackgroundPatient-reported outcome measures are increasingly used by clinicians to support communication in telephone- or face-to-face consultations with patients. A renal disease questionnaire has been developed, but not sufficiently evaluated through clinimetrics in clinical setting. Hence, we aimed to evaluate the content validity, construct validity and the test-retest reliability of a renal disease questionnaire to be used for clinical decision-making.MethodsA content, construct validity and test-retest reliability study was conducted in 3 nephrology outpatient clinics in Central Denmark Region, Denmark. Content validity (face validity, comprehensibility and relevance) was assessed among 8 patients and 6 clinicians. Reliability was assessed by asking outpatients with chronic kidney disease to complete the questionnaire twice. Reliability was assessed by kappa statistics and agreement by percentage. Construct validity was determined using 4 a priori defined hypotheses and comparing 2 known groups.ResultsFive new domains emerged, 6 items were rephrased and 3 items were removed following the content validity test. A total of 160 patients completed the questionnaire with median 8 days (IQR 2 days) between assessments. The test-retest reliability parameters of the single items in the questionnaire were substantial to almost perfect as all the observed weighted kappa values ranged from 0.61 to 0.91, 95% CI (0.34 to 0.95). In total, 61% of the single items showed almost perfect agreement. In total, 3 of the 4 hypotheses were accepted and 44% of the items showed satisfying known-group discriminative validity.ConclusionA renal disease questionnaire used for clinical decision-making in outpatient follow-up showed acceptable content validity and substantial to almost perfect reliability. Sufficient construct validity was not established. Incorporating the questionnaire into routine clinical practice may improve the evaluation of disease burden in patients with chronic kidney disease. We ask patients with chronic kidney disease (CKD) in Central Region Denmark to complete a questionnaire before each outpatient visit. The answers they provide are used to support communication with their health care provider. A questionnaire requires testing to ensure it can accurately capture important information about patient's symptoms and quality of life. When questionnaires are used to support communication between patients and health care professionals, they need to have good measurement properties. This means they need to be: (1) trustworthy, (2) relevant to a patient's health condition, (3) consistent and produce stable results every time. We explored the measurement properties of a questionnaire designed to be used in the face-to face outpatient visits for patients with CKD. We found that the questionnaire captured consistent and stable results. Using this questionnaire may help health care professionals to assess the patients´ burden of symptoms with a more patient-centered approach. Potentially, the use of the questionnaire will increase the patients´ ability to cope with their symptoms and strengthen patients´ involvement in the clinical decisions concerning their treatment.

Project description:Objectives/hypothesisTo explore possible factors that might impact a patient's choice to pursue endoscopic sinus surgery (ESS) or continue with medical management for treatment of refractory chronic rhinosinusitis (CRS).Study designCross-sectional evaluation of a multicenter prospective cohort.MethodsTwo hundred forty-two subjects with CRS were prospectively enrolled within four academic tertiary care centers across North America with ongoing symptoms despite prior medical treatment. Subjects either self-selected continued medical management (n = 62) or ESS (n = 180) for treatment of sinonasal symptoms. Differences in demographics, comorbid conditions, and clinical measures of disease severity between subject groups were compared. Validated metrics of social support, personality, risk aversion, and physician-patient relationships were compared using bivariate analyses, predicted probabilities, and receiver operating characteristic curves at the 0.05 alpha level.ResultsNo significant differences were found between treatment groups for any demographic characteristic, clinical cofactor, or measure of social support, personality, or the physician-patient relationship. Subjects electing to pursue sinus surgery did report significantly worse average quality-of-life (QOL) scores on the 22-item Sinonasal Outcome Test (SNOT-22; P < .001) compared to those electing continued medical therapy (54.6 ± 18.9 vs. 39.4 ± 17.7), regardless of surgical history or polyp status. SNOT-22 score significantly predicted treatment selection (odds ratio, 1.046; 95% confidence interval, 1.028-1.065; P < .001) and was found to accurately discriminate between subjects choosing endoscopic sinus surgery and those electing medical management 72% of the time.ConclusionsWorse patient-reported disease severity, as measured by the SNOT-22, was significantly associated with the treatment choice for CRS. Strong consideration should be given for incorporating CRS-specific QOL measures into routine clinical practice.Level of evidence2b.

Project description:BackgroundPoor patient-provider communication, among other reasons, is a notable barrier to contraceptive decision-making among Latinas. Patient-centered approaches to contraceptive counseling that optimize communication align with shared decision-making (SDM) -which is associated with satisfaction and continued contraceptive use among various populations.ObjectiveTo examine associations of patient-provider communication and importance of SDM tenets with consistent contraceptive use among a population of Latinas.Patient involvementFormative work for this study included prior qualitative and quantitative research with Latinas who expressed the importance of patient-provider communication during contraceptive counseling and therefore were instrumental in problem definition.MethodsCross-sectional surveys were administered to Latinas ages 15-29 years. Patient-provider communication, patient-reported importance of specific SDM tenets, and consistent contraception use were measured and analyzed for associations.Results103 Latinas (mean age = 21.4) participated. 33% of participants<21 years were using contraception consistently vs. 67% for those ≥ 21 (p = 0.003). Among participants ≥ 21, consistent users reported higher communication scores compared to inconsistent users and non-users (p = 0.042). For participants< 21, consistent users were more likely than inconsistent users and non-users to report that 2 SDM tenets (discussion of contraceptive preferences and avoidance of race/ethnic-based judgement) are important (p = 0.052, 0.028, respectively).DiscussionPatient-provider communication was especially important for Latinas ≥ 21 while using an SDM approach during counseling was highly valued by those<21. Patient-centered approaches to contraceptive counseling provide opportunities to optimize healthcare delivery for this vulnerable population.Practical valueResults from this research demonstrate that patient-centered communication is highly valued by Latina study participants and is an important consideration in their contraceptive counseling. Clinicians should consider employing techniques such as SDM as they seek to provide patient-centered care during contraceptive counseling for this patient population.

Project description:ImportanceHistorically, opioid pain medications have been overprescribed following thyroid and parathyroid surgery. Many narcotic prescriptions are incompletely consumed, creating waste and opportunities for abuse.ObjectiveTo determine whether limiting opioid prescriptions after outpatient thyroid and parathyroid surgery to patients who opt in to narcotic treatment reduces opioid consumption without increasing postoperative pain compared with usual care (routine narcotic prescriptions).Design, setting, and participantsA randomized clinical trial of Postoperative Opt-In Narcotic Treatment (POINT) or routine narcotic prescription (control) was conducted at a single tertiary referral center from June 1 to December 30, 2020. A total of 180 adults undergoing ambulatory cervical endocrine surgery, excluding patients currently receiving opioids, were assessed for eligibility. POINT patients received perioperative pain management counseling and were prescribed opioids only on patient request. Patients reported pain scores (0-10) and medication use through 7 daily postoperative surveys. Logistic regression was used to determine factors associated with opioid consumption.InterventionsPatients in the POINT group were able to opt in or out of receiving prescriptions for opioid pain medication on discharge. Control patients received routine opioid prescriptions on discharge.Main outcomes and measuresDaily peak pain score through postoperative day 7 was the primary outcome. Noninferiority was defined as a difference less than 2 on an 11-point numeric rating scale from 0 to 10. Analysis was conducted on the evaluable population.ResultsOf the 180 patients assessed for eligibility, the final study cohort comprised 102 patients: 48 randomized to POINT and 54 to control. Of these, 79 patients (77.5%) were women and median age was 52 (interquartile range, 43-62) years. A total of 550 opioid tablets were prescribed to the control group, and 230 tablets were prescribed to the POINT group, in which 23 patients (47.9%) opted in for an opioid prescription. None who opted out subsequently required rescue opioids. In the first postoperative week, 17 POINT patients (35.4% of survey responders in the POINT group) reported consuming opioids compared with 27 (50.0%) control patients (P = .16). Median peak outpatient pain scores were 6 (interquartile range, 4-8) in the control group vs 6 (interquartile range, 5-7) in the POINT group (P = .71). In multivariate analysis, patients with a history of narcotic use were 7.5 times more likely to opt in (95% CI, 1.61-50.11; P = .02) and 4.8 times more likely to consume opioids (95% CI, 1.04-1.52; P = .01). Higher body mass index (odds ratio, 1.11; 95% CI, 1.01-1.23; P = .03) and highest inpatient postoperative pain score (odds ratio, 1.24; 95% CI, 1.04-1.52; P = .02) were also associated with opioid consumption.Conclusions and relevanceIn this trial, an opt-in strategy for postoperative narcotics reduced opioid prescription without increasing pain after cervical endocrine surgery.Trial registrationClinicalTrials.gov Identifier: NCT04710069.

Dataset Information

Towards Patient-centered Decision-making in Breast Cancer Surgery: Machine Learning to Predict Individual Patient-reported Outcomes at 1-year Follow-up.

Objective

Summary of background data

Methods

Results

Conclusion

Publications

Towards Patient-centered Decision-making in Breast Cancer Surgery: Machine Learning to Predict Individual Patient-reported Outcomes at 1-year Follow-up.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets