Predicting functional outcomes in patients suffering from ischaemic stroke using initial admission variables and physiological data: a comparison between tree model and multivariate regression analysis

Cross-validation of predictive models for functional recovery after post-stroke rehabilitation

Background

Rehabilitation treatments and services are essential for the recovery of post-stroke patients’ functions; however, the increasing number of available therapies and the lack of consensus among outcome measures compromises the possibility to determine an appropriate level of evidence. Machine learning techniques for prognostic applications offer accurate and interpretable predictions, supporting the clinical decision for personalised treatment. The aim of this study is to develop and cross-validate predictive models for the functional prognosis of patients, highlighting the contributions of each predictor.

Methods

A dataset of 278 post-stroke patients was used for the prediction of the class transition, obtained from the modified Barthel Index. Four classification algorithms were cross-validated and compared. On the best performing model on the validation set, an analysis of predictors contribution was conducted.

Results

The Random Forest obtained the best overall results on the accuracy (76.2%), balanced accuracy (74.3%), sensitivity (0.80), and specificity (0.68). The combination of all the classification results on the test set, by weighted voting, reached 80.2% accuracy. The predictors analysis applied on the Support Vector Machine, showed that a good trunk control and communication level, and the absence of bedsores retain the major contribution in the prediction of a good functional outcome.

Conclusions

Despite a more comprehensive assessment of the patients is needed, this work paves the way for the implementation of solutions for clinical decision support in the rehabilitation of post-stroke patients. Indeed, offering good prognostic accuracies for class transition and patient-wise view of the predictors contributions, it might help in a personalised optimisation of the patients’ rehabilitation path.

Machine learning methods for functional recovery prediction and prognosis in post-stroke rehabilitation: a systematic review

BACKGROUND

Rehabilitation medicine is facing a new development phase thanks to a recent wave of rigorous clinical trials aimed at improving the scientific evidence of protocols. This phenomenon, combined with new trends in personalised medical therapies, is expected to change clinical practice dramatically. The emerging field of Rehabilomics is only possible if methodologies are based on biomedical data collection and analysis. In this framework, the objective of this work is to develop a systematic review of machine learning algorithms as solutions to predict motor functional recovery of post-stroke patients after treatment.

METHODS

We conducted a comprehensive search of five electronic databases using the Patient, Intervention, Comparison and Outcome (PICO) format. We extracted health conditions, population characteristics, outcome assessed, the method for feature extraction and selection, the algorithm used, and the validation approach. The methodological quality of included studies was assessed using the prediction model risk of bias assessment tool (PROBAST). A qualitative description of the characteristics of the included studies as well as a narrative data synthesis was performed.

RESULTS

A total of 19 primary studies were included. The predictors most frequently used belonged to the areas of demographic characteristics and stroke assessment through clinical examination. Regarding the methods, linear and logistic regressions were the most frequently used and cross-validation was the preferred validation approach.

CONCLUSIONS

We identified several methodological limitations: small sample sizes, a limited number of external validation approaches, and high heterogeneity among input and output variables. Although these elements prevented a quantitative comparison across models, we defined the most frequently used models given a specific outcome, providing useful indications for the application of more complex machine learning algorithms in rehabilitation medicine.

A closer examination of the interaction among risk factors for low back pain

PURPOSE

To evaluate the interactions of risk factors and identify their most powerful discrimination pathway for the occurrence of low back pain (LBP).

DESIGN

A cross-sectional study.

SETTING

Taiwan.

SUBJECTS

Taiwanese population of 30 to 64 years old.

MEASURES

A self-reported question, "Have you experienced LBP within the last 3 months?" was used to evaluate LBP. The study variables included demographics (age, gender, occupation, education level, marital status, and household income), biometric health measures (bone mineral density and body mass index), dietary habits (weekly milk, coffee, tea, and soybean consumption), and other lifestyle factors (smoking habits, alcohol consumption, betel nut chewing, body weight control, exercise regularity, and stress management).

ANALYSIS

Logistic regression and classification tree analyses.

RESULTS

A total of 969 Taiwanese participants were analyzed. Primary logistic regression analysis identified three critical risk factors (gender, bone mineral density, and exercise regularity) for the occurrence of LBP. By classification tree analysis, demographic factors, dietary habits, and lifestyle factors had modifying effects on LBP.

CONCLUSIONS

Various factors contribute to the risk of LBP. Interactions between risk factors should be considered when developing future strategies for the prevention and management of LBP.

Catechol-O-Methyltransferase Polymorphism Influences Outcome After Ischemic Stroke

Background. To explore whether a polymorphism in dopamine metabolism influences the effectiveness of neurological rehabilitation and the outcome after ischemic stroke. Methods. The Barthel Index (BI) and the Rivermead Motor Assessment (RMA) were assessed in 78 moderately affected stroke patients (1) after they had entered a neurological inpatient rehabilitation, (2) after 4 weeks of rehabilitation therapy, and (3) 6 months later. Polymorphisms of the gene encoding catechol- O-methyltransferase (COMT) were determined. BI and RMA results were analyzed with respect to the genetic profiles of COMT. Results. Carriers of COMT Val/Val alleles showed better results in BI and RMA than COMT Met/Met carriers at all 3 time points. Val/Met carriers exhibited results in between the homozygotes, suggesting a gene–dose relationship. Altogether, BI and RMA results were highly correlated. Conclusion. Stroke patients with COMT Val/Val alleles had higher motor functions and abilities of activities of daily living even at the beginning of the rehabilitation period. All patient groups improved during the rehabilitation period to a similar degree, suggesting that physical therapy is comparably effective in all polymorphism subtypes.

Risk Factor of Visuospatial Neglect: A Study of Association Between Visuospatial Neglect and Anemia

Objective

To investigate the correlation between visuospatial neglect and anemia in patients with right cerebral infarction, as well as to identify the risk factor of neglect and furnish preliminary data on rehabilitation management.

Methods

The line bisection test and Albert test were conducted on subjects with right cerebral infarction in order to analyze neglect severity. Multiple linear regression analysis was conducted to investigate correlation between neglect severity and hemoglobin and hematocrit level. Logistic regression analysis was applied to identify the risk factor of neglect.

Results

Visuospatial neglect was observed in 33 subjects out of 124. Hemoglobin and hematocrit were not directly correlated with visuospatial neglect severity, whereas infarct size was directly correlated. Subjects with visuospatial neglect were characterized by a large infarct size, a low score in the Mini-Mental State Examination and long hospital stay.

Conclusion

In this study, visuospatial neglect was found to be uncorrelated with anemia. It implies that emphasis should be placed on the early detection of anemia and neglect in patients with left hemiplegia, the formulation of respective therapeutic plans and improvement of prognosis. The study found that the possibility of a visuospatial neglect occurrence increases with infarct size. In this regard, it is required that visuospatial neglect was detected and treated in the earliest possible stage, notwithstanding the difficulty that lies in the precise measurement of the severity.

Effects of repetitive hyperbaric oxygen treatment in patients with acute cerebral infarction: a pilot study

The role of hyperbaric oxygen therapy (HBOT) in the treatment of acute ischemic stroke is controversial. This prospective study assessed the efficacy and safety of HBOT as adjuvant treatment on 46 acute ischemic stroke in patients who did not receive thrombolytic therapy. The HBOT group (n = 16) received conventional medical treatment with 10 sessions of adjunctive HBOT within 3-5 days after stroke onset, while the control group (n = 30) received the same treatment but without HBOT. Early (around two weeks after onset) and late (one month after onset) outcomes (National Institutes of Health Stroke Scale, NIHSS scores) and efficacy (changes of NIHSS scores) of HBOT were evaluated. The baseline clinical characteristics were similar in both groups. Both early and late outcomes of the HBOT group showed significant difference (P ≤ 0.001). In the control group, there was only significant difference in early outcome (P = 0.004). For early efficacy, there was no difference when comparing changes of NIHSS scores between the two groups (P = 0.140) but there was statistically significant difference when comparing changes of NIHSS scores at one month (P ≤ 0.001). The HBOT used in this study may be effective for patients with acute ischemic stroke and is a safe and harmless adjunctive treatment.

Prognostic significance of smoking in patients with acute ischemic stroke within 3 months of onset

BACKGROUND

Various factors that have been implicated in recovery after the acute phase of stroke have not been well evaluated.

METHODS

To identify prognostic factors affecting outcomes at 90 days after stroke from the viewpoint of recovery patterns, we enrolled 660 patients from the Edaravone and Argatroban Stroke Therapy for Acute Ischemic Stroke study database. Fourteen groups of patients were identified based on an analysis of their recovery patterns according to changes in their National Institutes of Health Stroke Scale scores during the first 21 days. These groups were then divided into 2 groups: favorable recovery trend (patterns 1-3; n = 486) and poor recovery trend (patterns 4-14; n = 174). Patterns with >80% of the patients experiencing a favorable outcome (National Institutes of Health Stroke Scale score of ≤ 4 at 90 days) were defined as the favorable recovery trend group, whereas patterns that included ≤ 80% favorable outcomes were defined as the poor recovery trend group.

RESULTS

Using the poor recovery trend group, logistic regression analysis found that after controlling for covariates, lower scores at admission, fewer ischemic lesions, and nonsmoking were significant prognostic factors for a favorable outcome at 90 days.

CONCLUSIONS

Based on a detailed analysis of the relationship between recovery patterns after stroke and clinical outcomes in the chronic stage of stroke, smoking cessation may improve the prognosis of patients after stroke.

Factors related to daily life interference in lung cancer patients: a cross-sectional regression tree study

PURPOSE

To identify the symptom combination patterns and symptom severity levels that induce severe symptom interference in daily life activities, including physical and psychological activity interference in lung cancer patients.

METHODS

In a cross-sectional descriptive study using convenience sampling, 131 participants were recruited at a medical center in northern Taiwan. The Eastern Cooperative Oncology Group (ECOG) performance status was used to assess performance status, and the Taiwanese version of the M.D. Anderson Symptom Inventory (MDASI-T) was used to assess symptom severity and symptom interference in daily life activities including physical and psychological activities. Regression tree models were applied to examine variable combinations for symptom interference level in daily life activities, including physical and psychological activity interference.

RESULTS

Study results revealed that the performance status is the key discriminator of the symptom interference level in daily life and physical activities, but distress severity is the key discriminating factor of the symptom interference level in psychological activities. The performance status and distress severity, plus other factors, further specifically show the discrimination paths and interactions between the risk groups.

CONCLUSIONS

This study provided an alternative approach to identify low- and high-risk groups of symptom interference among lung cancer patients in Taiwan. Increased awareness and further understanding of the risk combinations and discriminate levels of symptom severity that induced high symptom interference offer different perspectives to develop patient-centered care planning for lung cancer patient rehabilitation.

The prognostic roles of initial glucose level and functional outcomes in patients with ischemic stroke: difference between diabetic and nondiabetic patients

PURPOSE

To identify the prognostic roles of initial glucose level in the emergency department among patients with acute ischemic stroke, and to evaluate whether or not the impact varied with diabetes.

METHODS

A total of 774 first-ever ischemic stroke patients admitted to the emergency department of a tertiary hospital in Taiwan. Participants were stratified into tertiles by initial glucose level in the emergency department. The Barthel Index (BI) and modified Rankin Scale (mRS) were used to assess functional outcome 3 months after the ischemic stroke.

RESULTS

Initial glucose level was significantly associated with functional outcome 3 months after ischemic stroke. Compared with the 1st tertile, adjusted odds ratios for 2nd and 3rd tertiles of initial glucose level were 1.84 (95% confidence interval [CI], 1.15-2.95) and 4.57 (95% CI, 2.87-7.29), for BI of less than 60, 2.07 (95% CI, 1.28-3.36) and 4.45 (95% CI, 2.76-7.18) for mRS score of >2. The association was apparent among the nondiabetic stroke patients compared with diabetic patients.

CONCLUSIONS

A positive association was found between initial glucose level and poor functional outcome at 3 months in patients with acute ischemic stroke, particularly among nondiabetic patients.