Predicting Survival Outcomes in the Presence of Unlabeled Data

Haredasht, Fateme Nateghi; Vens, Celine

doi:10.1007/s10994-022-06257-x

Cited by 3 publications

(1 citation statement)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For all ML models, the same folds partitioning is used in order to guarantee fair comparisons. Also, it is worthwhile to mention that due to the small sample size and high imbalance between classes, no hyperparameter tuning has been employed and the default hyperparameter values for each ML model are used instead, following the approach proposed in previous studies 32 , 62 . Notwithstanding, cost sensitive learning is applied to deal with class imbalance by re-weighting the loss function toward the less represented (i.e., minority) class 63 .…”

Section: Experimental Set-upmentioning

confidence: 99%

Causal impact evaluation of occupational safety policies on firms’ default using machine learning uplift modelling

Barile,

Forti,

Marrocco

et al. 2024

Sci Rep

View full text Add to dashboard Cite

It is often undermined that occupational safety policies do not only displace a direct effect on work well-being, but also an indirect effect on firms’ economic performances. In such context, econometric models dominated the scenes of causality until recently while Machine Learning models were seen with skepticism. With the rise of complex datasets, an ever-increasing need for automated algorithms capable to handle complex non-linear relationships between variables has brought to uncover the power of Machine Learning for causality. In this paper, we carry out an evaluation of a public aid-scheme implemented in Italy and oriented to support investment of small and medium enterprises (SMEs) in occupational safety and health (OSH) for assessing the impact on the survival of corporations. A comparison of thirteen models is performed and the Individual Treatment Effect (ITE) estimated and validated based on the AUUC and Qini score for which best values of 0.064 and 0.407, respectively, are obtained based on the Light Gradient Boosting Machine (LightGBM). An additional in-depth statistical analysis also revealed that the best beneficiaries of the policy intervention are those firms that experience performance issues in the period just before the interventions and for which the increased liquidity brought by the policy may have prevented default.

show abstract

Section: Experimental Set-upmentioning

confidence: 99%

Causal impact evaluation of occupational safety policies on firms’ default using machine learning uplift modelling

Barile,

Forti,

Marrocco

et al. 2024

Sci Rep

View full text Add to dashboard Cite

show abstract

Exploiting Censored Information in Self-Training for Time-to-Event Prediction

Haredasht,

Dauda,

Vens

2023

IEEE Access

Self Cite

View full text Add to dashboard Cite

A common problem in medical applications is predicting the time until an event of interest such as the onset of a disease, time to tumor recurrence, and time to mortality. Traditionally, classical survival analysis techniques have been used to address this problem. However, these techniques are of limited usage when considering nonlinear and interaction effects among biomarkers, and high profiling survival datasets. Although supervised machine learning techniques have shown some advantages over standard statistical methods in handling high-dimensional datasets, their application to survival analysis is at best limited. A major reason behind this is the difficulty in processing censored data, which is a common component of survival analysis. In this paper, we have transformed the time-to-event prediction problem into a semisupervised regression problem in which we use a self-training wrapper approach with random survival forests as the base learner. In this approach, censored observations are introduced as partially labeled observations since their predicted time (target value) should exceed the censoring time. First, the algorithm builds a base model over the observed instances and then augments them iteratively with highly confident predictions over the censored set, using a smart stopping criterion based on the censoring time. The proposed approach has been evaluated and compared on fifteen real-world survival analysis datasets, including clinical and highdimensional data. The ability of our proposed approach to integrate partial supervision information within a semi-supervised learning strategy has enabled it to achieve competitive performance compared to baseline models, particularly in the case of a high-dimensional regime.INDEX TERMS Random survival forest, Self-training, Semi-supervised learning, Survival analysis.

show abstract

Validated risk prediction models for outcomes of acute kidney injury: a systematic review

et al. 2023

Self Cite

View full text Add to dashboard Cite

Background Acute Kidney Injury (AKI) is frequently seen in hospitalized and critically ill patients. Studies have shown that AKI is a risk factor for the development of acute kidney disease (AKD), chronic kidney disease (CKD), and mortality. Methods A systematic review is performed on validated risk prediction models for developing poor renal outcomes after AKI scenarios. Medline, EMBASE, Cochrane, and Web of Science were searched for articles that developed or validated a prediction model. Moreover, studies that report prediction models for recovery after AKI also have been included. This review was registered with PROSPERO (CRD42022303197). Result We screened 25,812 potentially relevant abstracts. Among the 149 remaining articles in the first selection, eight met the inclusion criteria. All of the included models developed more than one prediction model with different variables. The models included between 3 and 28 independent variables and c-statistics ranged from 0.55 to 1. Conclusion Few validated risk prediction models targeting the development of renal insufficiency after experiencing AKI have been developed, most of which are based on simple statistical or machine learning models. While some of these models have been externally validated, none of these models are available in a way that can be used or evaluated in a clinical setting.

show abstract

Predicting Survival Outcomes in the Presence of Unlabeled Data

Cited by 3 publications

References 30 publications

Causal impact evaluation of occupational safety policies on firms’ default using machine learning uplift modelling

Causal impact evaluation of occupational safety policies on firms’ default using machine learning uplift modelling

Exploiting Censored Information in Self-Training for Time-to-Event Prediction

Validated risk prediction models for outcomes of acute kidney injury: a systematic review

Contact Info

Product

Resources

About