Predictability of the future development of aggressive behavior of cranial dural arteriovenous fistulas based on decision tree analysis

Risk Factors of Aggressive Clinical Presentation in Patients with Angiographically Aggressive Cranial Dural Arteriovenous Fistulas

Compared to nonaggressive cranial dural arteriovenous fistulae (cDAVF), aggressive cDAVF carries leptomeningeal venous drainage (LVD) and has approximately 15% annual risk of hemorrhagic and non-hemorrhagic aggressive neurological presentations. In terms of aggressive clinical presentations, the previous classification does not adequately differentiate the higher risk group from the lower risk group. Herein, we retrospectively collected a series of patients with aggressive cDAVF and explored the risk factors for differentiating the higher-risk group from the lower-risk group with aggressive clinical presentations. We retrospectively collected patients with aggressive cDAVF from March 2011 to March 2019. The risk of aggressive clinical presentation was recorded. Risk factors were included in the analysis for aggressive clinical presentations. From March 2011 to March 2019, 37 patients had aggressive cDAVF. Among them, 24 presented with aggressive clinical presentation (20, hemorrhagic presentation; four, non-hemorrhagic presentation). In patients presenting with hemorrhage, four patients experienced early rebleeding after diagnosis. In the univariate analysis, risk location, directness of LVD, exclusiveness of LVD, and venous strain were significantly different in patients with aggressive clinical presentation. In the multivariate analysis, exclusiveness of LVD and venous strain were observed, with a significant difference between patients with aggressive clinical presentation and those with benign clinical presentation. Among patients with angiographically aggressive cDAVFs, approximately 65% presented with aggressive clinical presentations in our series. Among all potential risk factors, patients with exclusiveness of LVD and venous strain have even higher risk and should be treated aggressively and urgently.

Impact of subcallosal artery origin and A1 asymmetry on surgical outcomes of anterior communicating artery aneurysms

BACKGROUND

Surgical clipping of anterior communicating artery (ACoA) aneurysms remains challenging due to their complex anatomy. Anatomical risk factors for ACoA aneurysm surgery require further elucidation. The aim of this study is to investigate whether proximity of the midline perforating artery, subcallosal artery (SubCA), and associated anomaly of the ACoA complex affect functional outcomes of ACoA aneurysm surgery.

METHODS

A total of 92 patients with both unruptured and ruptured ACoA aneurysms, who underwent surgical clipping, were retrospectively analyzed from a multicenter, observational cohort database. Association of ACoA anatomy with SubCA origin at the aneurysmal neck under microsurgical observation was analyzed in the interhemispheric approach subgroup (n = 56). Then, we evaluated whether anatomical factors associated with SubCA neck origin affected surgical outcomes in the entire cohort (both interhemispheric and pterional approaches, n = 92).

RESULTS

In the interhemispheric approach cohort, combination of A1 asymmetry and aneurysmal size ≥ 5.0 mm was stratified to have the highest probability of the SubCA neck origin by a decision tree analysis. Then, among the entire cohort using either interhemispheric or pterional approach, combination of A1 asymmetry and aneurysmal size ≥ 5.0 mm was significantly associated with poor functional outcomes by multivariable logistic regression analysis (OR 6.76; 95% CI 1.19-38.5; p = 0.03) as compared with A1 symmetry group in the acute subarachnoid hemorrhage settings.

CONCLUSION

Combination of A1 asymmetry and larger aneurysmal size was significantly associated with SubCA aneurysmal neck origin and poor functional outcomes in ACoA aneurysm surgery. Interhemispheric approach may be proposed to provide a wider and unobstructed view of SubCA for ACoA aneurysms with this high-risk anatomical variant.

Relative signal intensity on time-of-flight magnetic resonance angiography as a novel indicator of aggressive presentation of intracranial dural arteriovenous fistulas

Asymptomatic dural arteriovenous fistulas (DAVFs) with cortical venous reflux (CVR) are now more commonly encountered. However, patients with an incidental onset may have a less aggressive clinical course. It is desirable to explore methods and indicators to predict the clinical outcomes. This study investigates whether the relative signal intensity (rSI) of the draining vessels on the time-of-flight magnetic resonance angiography is related to clinical behavior in patients with intracranial DAVFs. We retrospectively reviewed 36 intracranial DAVFs. The patients were categorized as those with either aggressive-presentation or non-aggressive-presentation (n = 16 and 20, respectively). The rSIs of the shunt points, affected sinuses, and veins with CVR were compared between the two groups. The two groups were not significantly different in terms of rSIs of the shunt points and affected sinuses (p = 0.37 and 0.41, respectively). However, a significant positive correlation was observed in the rSI of the veins with CVR between the aggressive and non-aggressive behavior groups (p < 0.0001). The rSI of the veins with CVR could serve as a reliable indicator of aggressive behavior in intracranial DAVFs, and its optimal cutoff value was 1.63 with high sensitivity and specificity for predicting aggressive behavior (area under the curve, 0.909).

Predictability of postoperative recurrence on hepatocellular carcinoma through data mining method

Hepatocellular carcinoma (HCC) is a highly lethal tumor and the majority of postoperative patients experience recurrence. In the present study, we focus on the predictability of postoperative recurrence on HCC through the data mining method. In total, 323 patients with HCC who underwent hepatic resection were included in the present study, 156 of whom suffered from cancer recurrence. Clinicopathological data including prognosis were analyzed using the data mining method for the predictability of postoperative recurrence on HCC. The resulting alternating decision tree (ADT) was described using data mining method. This tree was validated using a 10-fold cross validation process. The average and standard deviation of the accuracy, sensitivity, and specificity were 69.0±8.2, 59.7±14.5 and 77.7±10.2%, respectively. The identified postoperative recurrence factors were age, viral hepatitis, stage, GOT and T-cholesterol. Data mining method could identify the factors associated at different levels of significance with postoperative recurrence of HCC. These factors could help to predict the postoperative recurrence of HCC.

Decision tree analysis in subarachnoid hemorrhage: prediction of outcome parameters during the course of aneurysmal subarachnoid hemorrhage using decision tree analysis

OBJECTIVEThe aim of this study was to create prediction models for outcome parameters by decision tree analysis based on clinical and laboratory data in patients with aneurysmal subarachnoid hemorrhage (aSAH).METHODSThe database consisted of clinical and laboratory parameters of 548 patients with aSAH who were admitted to the Neurocritical Care Unit, University Hospital Zurich. To examine the model performance, the cohort was randomly divided into a derivation cohort (60% [n = 329]; training data set) and a validation cohort (40% [n = 219]; test data set). The classification and regression tree prediction algorithm was applied to predict death, functional outcome, and ventriculoperitoneal (VP) shunt dependency. Chi-square automatic interaction detection was applied to predict delayed cerebral infarction on days 1, 3, and 7.RESULTSThe overall mortality was 18.4%. The accuracy of the decision tree models was good for survival on day 1 and favorable functional outcome at all time points, with a difference between the training and test data sets of < 5%. Prediction accuracy for survival on day 1 was 75.2%. The most important differentiating factor was the interleukin-6 (IL-6) level on day 1. Favorable functional outcome, defined as Glasgow Outcome Scale scores of 4 and 5, was observed in 68.6% of patients. Favorable functional outcome at all time points had a prediction accuracy of 71.1% in the training data set, with procalcitonin on day 1 being the most important differentiating factor at all time points. A total of 148 patients (27%) developed VP shunt dependency. The most important differentiating factor was hyperglycemia on admission.CONCLUSIONSThe multiple variable analysis capability of decision trees enables exploration of dependent variables in the context of multiple changing influences over the course of an illness. The decision tree currently generated increases awareness of the early systemic stress response, which is seemingly pertinent for prognostication.

Early rebleeding of intracranial dural arteriovenous fistulas after an intracranial hemorrhage

BACKGROUND

The aim of this study was to evaluate the clinical and angiographic characteristics of dural arteriovenous fistulas (DAVF) presenting with intracranial hemorrhage (ICH), with a focus on early rebleeding according to the initial hemorrhage type.

METHOD

The clinical and radiologic features of 21 dAVFs that presented with intracranial hemorrhage were retrospectively reviewed. The hemorrhage type was classified as pure intraparenchymal hemorrhage (pIPH) and subarachnoid or subdural hemorrhage with IPH (non-pIPH).

RESULTS

There were 13 patients with pIPH and 8 with non-pIPH. The median follow-up period was 30 months (range, 1-116 months), and the median interval from hemorrhage to treatment was 4 days (range, 0-72 days). Rebleeding occurred in 8 (38.1%) of 21 patients. Four (50%) of eight patients with non-pIPH suffered from early rebleeding within 3 days, while there was no early rebleeding in patients with pIPH. There was a significantly higher rate of early rebleeding in the non-pIPH group (p = 0.012). Angiographically, venous ectasia (p = 0.005) and direct cortical venous drainage (dCVD) (p = 0.008) showed a significantly higher proportion in the non-pIPH group than in the pIPH group.

CONCLUSIONS

DAVFs with ICH is likely to rebleed after the first hemorrhage. Thus, early treatment can be needed in all DAVFs with ICH. In addition, DAVFs that presenting with non-pIPH and containing venous ectasia or dCVD on initial angiography may have a higher risk of early rebleeding. Therefore, cautious attention and urgent treatment are necessary for these patients.

Predictability of Pressure Ulcers Based on Operation Duration, Transfer Activity, and Body Mass Index Through the Use of an Alternating Decision Tree

OBJECTIVE

To develop a prediction model for pressure ulcer cases that continue to occur at an acute care hospital with a low occurrence rate of pressure ulcers.

METHODS

Analyzing data were collected from patients hospitalized at Tokushima University Hospital during 2012 using an alternating decision tree (ADT) data mining method.

RESULTS

The ADT-based analysis revealed transfer activity, operation time, and low body mass index (BMI) as important factors for predicting pressure ulcer development.

DISCUSSION

Among the factors identified, only "transfer activity" can be modified by nursing intervention. While shear force and friction are known to lead to pressure ulcers, transfer activity has not been identified as such. Our results suggest that transfer activities creating shear force and friction correlate with pressure ulcer development. The ADT algorithm was effective in determining prediction factors, especially for highly imbalanced data. Our three stumps ADT yielded accuracy, sensitivity, and specificity values of 72.1%±3.7%, 79.3%±18.1%, and 72.1%±3.8%, respectively.

CONCLUSION

Transfer activity, identified as an interventional factor, can be modified through nursing interventions to prevent pressure ulcer formation. The ADT method was effective in identifying factors within largely imbalanced data. J. Med. Invest. 63: 248-255, August, 2016.

Systemic inflammation and family history in relation to the prevalence of type 2 diabetes based on an alternating decision tree

To investigate unknown patterns associated with type 2 diabetes in the Japanese population, we first used an alternating decision tree (ADTree) algorithm, a powerful classification algorithm from data mining, for the data from 1,102 subjects aged 35-69 years. On the basis of the investigated patterns, we then evaluated the associations of serum high-sensitivity C-reactive protein (hs-CRP) as a biomarker of systemic inflammation and family history of diabetes (negative, positive or unknown) with the prevalence of type 2 diabetes because their detailed associations have been scarcely reported. Elevated serum hs-CRP levels were proportionally associated with the increased prevalence of type 2 diabetes after adjusting for probable covariates, including body mass index and family history of diabetes (P for trend = 0.016). Stratified analyses revealed that elevated serum hs-CRP levels were proportionally associated with increased prevalence of diabetes in subjects without a family history of diabetes (P for trend = 0.020) but not in those with a family history or with an unknown family history of diabetes. Our study demonstrates that systemic inflammation was proportionally associated with increased prevalence of type 2 diabetes even after adjusting for body mass index, especially in subjects without a family history of diabetes.

Guidelines for Developing and Reporting Machine Learning Predictive Models in Biomedical Research: A Multidisciplinary View

Background

As more and more researchers are turning to big data for new opportunities of biomedical discoveries, machine learning models, as the backbone of big data analysis, are mentioned more often in biomedical journals. However, owing to the inherent complexity of machine learning methods, they are prone to misuse. Because of the flexibility in specifying machine learning models, the results are often insufficiently reported in research articles, hindering reliable assessment of model validity and consistent interpretation of model outputs.

Objective

To attain a set of guidelines on the use of machine learning predictive models within clinical settings to make sure the models are correctly applied and sufficiently reported so that true discoveries can be distinguished from random coincidence.

Methods

A multidisciplinary panel of machine learning experts, clinicians, and traditional statisticians were interviewed, using an iterative process in accordance with the Delphi method.

Results

The process produced a set of guidelines that consists of (1) a list of reporting items to be included in a research article and (2) a set of practical sequential steps for developing predictive models.

Conclusions

A set of guidelines was generated to enable correct application of machine learning models and consistent reporting of model specifications and results in biomedical research. We believe that such guidelines will accelerate the adoption of big data analysis, particularly with machine learning methods, in the biomedical research community.

Risk of Complications after Thyroidectomy and Parathyroidectomy: A Case Series with Planned Chart Review

OBJECTIVE

To develop a predictive model for the risk of complications after thyroid and parathyroid surgery.

STUDY DESIGN

Case series with planned chart review of patients undergoing surgery, 2007-2013.

SETTING

Kaiser Permanente Northern California and Kaiser Permanente Southern California.

SUBJECTS AND METHODS

Patients (N = 16,458) undergoing thyroid and parathyroid procedures were randomly assigned to model development and validation groups. We used univariate analysis to assess relationships between each of 28 predictor variables and 30-day complication rates. We subsequently entered all variables into a recursive partitioning decision tree analysis, with P < .05 as the basis for branching.

RESULTS

Among patients undergoing thyroidectomies, the most important predictor variable was thyroid cancer. For patients with thyroid cancer, additional risk predictors included coronary artery disease and central neck dissection. For patients without thyroid cancer, additional predictors included coronary artery disease, dyspnea, complete thyroidectomy, and lobe size. Among patients undergoing parathyroidectomies, the most important predictor variable was coronary artery disease, followed by cerebrovascular disease and chronic kidney disease. The model performed similarly in the validation groups.

CONCLUSION

For patients undergoing thyroid surgery, 7 of 28 predictor variables accounted for statistically significant differences in the risk of 30-day complications; for patients undergoing parathyroid surgery, 3 variables accounted for significant differences in risk. This study forms the foundation of a parsimonious model to predict the risk of complications among patients undergoing thyroid and parathyroid surgery.