Hadi Akbarzadeh Khorshidi

Background: Several genomic risk stratification tests are available to predict the risk of metastasis and mortality for prostate cancer patients at the time of diagnosis. However, the evidence supporting the clinical utility of genomic risk stratification tools is fragmented, posing challenges in assessing their real‐world clinical impact and cost‐effectiveness. Objective: To review and summarise the clinical impact and health economic evidence of the four types of genomic risk stratification tests and to validate their clinical impact and health economic evaluation outcomes. Method: A systematic search was conducted in the Scopus, EMB Reviews and Google Scholar databases. Eligible publications were selected based on the eligible patient cohort, genomic test, initial NCCN risk categories, the clinical impact of the genomic test and health evaluation outcomes. Conclusion: 26 clinical impact evidence studies and four health economic evaluation studies were included. Most clinical studies indicated that genomic tests reclassified patients' risk predictions into both lower‐ and higher‐risk groups. The reclassification outcomes influenced patients' treatment decisions between active surveillance and radical treatment. The prognostic value of the genomic tests was validated in terms of biopsy upgrade, metastasis and death. The limited number of health economic studies reported that the Oncotype DX Prostate Score and ProMark were cost‐effective, while the Prolaris was cost‐saving in the US but not in Canada. The evaluation of the Decipher Genomic Classifier at the time of diagnosis was not available. More long‐term clinical evidence is needed, as are updated health economic evaluations, to determine the cost‐effectiveness of integrating genomic risk stratification into prostate cancer treatment decision‐making in clinical practice.

Treatment-induced ovarian function loss is a significant concern for many young patients with breast cancer. Accurately predicting this risk is crucial for counselling young patients and informing their fertility-related decision-making. However, current risk prediction models for treatment-related ovarian function loss have limitations. To provide a broader representation of patient cohorts and improve feature selection, we combined retrospective data from six datasets within the FoRECAsT (Infertility after Cancer Predictor) databank, including 2679 pre-menopausal women diagnosed with breast cancer. This combined dataset presented notable missingness, prompting us to employ cross imputation using the k-nearest neighbours (KNN) machine learning (ML) algorithm. Employing Lasso regression, we developed an ML model to forecast the risk of treatment-related amenorrhea as a surrogate marker of ovarian function loss at 12 months after starting chemotherapy. Our model identified 20 variables significantly associated with risk of developing amenorrhea. Internal validation resulted in an area under the receiver operating characteristic curve (AUC) of 0.820 (95% CI: 0.817–0.823), while external validation with another dataset demonstrated an AUC of 0.743 (95% CI: 0.666–0.818). A cutoff of 0.20 was chosen to achieve higher sensitivity in validation, as false negatives—patients incorrectly classified as likely to regain menses—could miss timely opportunities for fertility preservation if desired. At this threshold, internal validation yielded sensitivity and precision rates of 91.3% and 61.7%, respectively, while external validation showed 92.9% and 60.0%. Leveraging ML methodologies, we not only devised a model for personalised risk prediction of amenorrhea, demonstrating substantial enhancements over existing models but also showcased a robust framework for maximally harnessing available data sources.

Treatment-induced ovarian function loss is a significant concern for many young patients with breast cancer. Accurately predicting this risk is crucial for counselling young patients and informing their fertility-related decision-making. However, current risk-prediction models for treatment-related ovarian function loss have limitations. To provide a broader representation of patient cohorts and improve feature selection, we combined retrospective data from six datasets within the FoRECAsT (Infertility after Cancer Predictor) databank, including 2,679 pre-menopausal women diagnosed with breast cancer. This combined dataset presented notable missingness, prompting us to employ cross imputation using the k-nearest neighbours (KNN) machine learning (ML) algorithm. Employing Lasso regression, we developed an ML model to forecast the risk of treatment-related amenorrhea as a surrogate marker of ovarian function loss at 12 months after starting chemotherapy. Our model identified 20 variables significantly associated with risk of developing amenorrhea. Internal validation resulted in an area under the receiver operating characteristic curve (AUC) of 0.820 (95% CI: 0.817-0.823), while external validation with another dataset demonstrated an AUC of 0.743 (95% CI: 0.666-0.818). A cutoff of 0.20 was chosen to achieve higher sensitivity in validation, as false negatives—patients incorrectly classified as likely to regain menses—could miss timely opportunities for fertility preservation if desired. At this threshold, internal validation yielded sensitivity and precision rates of 91.3% and 61.7%, respectively, while external validation showed 92.9% and 60.0%. Leveraging ML methodologies, we not only devised a model for personalised risk prediction of amenorrhea, demonstrating substantial enhancements over existing models but also showcased a robust framework for maximally harnessing available data sources.

Precision oncology using biomarkers and genomic testing facilitates individualised diagnosis and consequently treatment for cancer patients. Standard of Care is conducted through assays of biomarkers for treatment selection. However, Whole Genome Sequencing (WGS) provides excessively more information about actionable mutations and complex interventions. Despite its potentials in cancer management, WGS is restricted by its high cost and limited capacity. Implementing WGS is complex because it influences several elements of the care delivery and deals with the dynamic changes in supply, demand, technology, and cost. In this study, we use a System Dynamics (SD) model to evaluate WGS implementation. We also use the developed SD model to conduct sensitivity analysis to derive insights for decision-making on genomic testing scenarios. Various scenarios have been compared by the number of diagnosed patients and total cost. These scenarios have been created through alterations in WGS referral system, capacity, price, and willingness rate as well as testing time across different hospitals. The results showed that referring more patients for WGS will increase the number of diagnosed patients (by 30%), however it will impose more cost on the system (by 20%) due to higher cost of WGS. Therefore, investment strategies toward reducing cost of WGS as well as reimbursement policies can be beneficial to make WGS more economic and accessible. This study shows how the developed SD model can contribute to providing insights for policy makers in finding the optimal way for implementing WGS.

Class imbalance is a substantial challenge in classifying many real-world cases. Synthetic over-sampling methods have been effective to improve the performance of classifiers for imbalance problems. However, most synthetic over-sampling methods generate synthetic instances within the convex hull formed by the existing minority instances as they only concentrate on the minority class and ignore the vast information provided by the majority class. They also often do not perform well for extremely imbalanced data, as fewer minority instances mean less information with which to generate synthetic instances. Moreover, existing methods that generate synthetic instances using the majority class distributional information cannot perform effectively when the majority class has a multi-modal distribution. We propose a new method to generate diverse and adaptable synthetic instances using Synthetic Over-sampling with Minority and Majority classes (SOMM). SOMM generates synthetic instances diversely within the minority data space. It updates the generated instances adaptively to the neighbourhood including both classes. Thus, SOMM performs well for imbalance problems. We examine the performance of SOMM for binary multiclass imbalance classification problems for different imbalance levels. The empirical results and nonparametric statistical testing show the superiority of SOMM compared to existing methods. We also discuss the strengths and limitations of SOMM through visualisations.

Background: Diabetes is a rapidly growing global health issue, with the hidden burden of undiagnosed cases leading to severe complications and escalating health care costs. Methods: This study investigated the potential of integrated behavioral frameworks to predict health-seeking behaviors and improve diabetes diagnosis timelines through the development of an agent-based model. Focusing on Narromine and Gilgandra in New South Wales, Australia, the model captured the integrative influence of 3 social theories—theory of planned behavior (TPB), health belief model (HBM), and goal framing theory (GFT)—on health care decisions across behavioral and nonbehavioral variables, providing a robust analysis of temporal diagnostic patterns, health care utilization, and costs. Results: Our comparative experiments indicated that this multitheory framework improved predictive accuracy by 15% to 30% compared with single-theory models, effectively capturing the interplay of planned, belief-driven, and context-based health behaviors. Spatial-temporal analysis highlighted key regional and demographic variations in diagnosis behaviors. While early, planned medical visits were prevalent in regions with better access (Gilgandra), areas with limited infrastructure saw a reliance on hospital-based diagnoses (Narromine). Health care cost analysis demonstrated a nonlinear expenditure pattern, suggesting that these theories defy conventional linear cost trends. Scenario analysis demonstrated the impact of targeted interventions. Gender-specific awareness initiatives in Gilgandra reduced late-diagnosis rates among men by approximately 15%, while enhanced access to care in Narromine decreased hospital-based late diagnoses from a baseline of 80% to around 60%. Conclusions: This study contributes an empirically grounded, policy-oriented decision support tool to inform targeted interventions, offering novel insights to improve diabetes management. Highlights: We explored the delay in diabetes diagnosis, particularly within remote Australian communities, through looking into the health care–seeking behavior of individuals displaying diabetes symptoms. We developed an innovative agent-based model to craft a dynamic decision support tool for policy makers by providing unique insights into the health behaviors of diabetes patients. Our study contributes significantly to the understanding of public health management with particular concerns around diabetes, as well as equips the New South Wales Ministry of Health with impactful insights into the consequences of their decisions.

Background Cancer screening plays a critical role in early disease detection and improving outcomes. In Australia, established screening protocols for colorectal, breast and cervical cancer have significantly contributed to timely cancer detection. However, the recent introduction of multi-cancer early detection (MCED) tests arguably can disrupt current screening, yet the extent to which these tests provide additional benefits remains uncertain. We present the development and initial validation of a system dynamics (SD) model that estimates the additional cancer detections and costs associated with MCED tests. Aim This article describes the development of a simulation model built to evaluate the additional patient diagnoses and the economic impact of incorporating MCED testing alongside Australia’s well-established standard of care (SOC) screening programs for colorectal, breast, cervical and lung cancers. The model was designed to estimate the additional number of patients diagnosed at each cancer stage (stage I, II, III, IV, or unknown) and the associated costs. This simulation model allows for the analysis of multiple scenarios under a plausible set of assumptions regarding population-level participation rates. Methods An SD model was developed to represent the existing SOC national cancer screening pathways and to integrate potential clinical pathways that could be introduced by MCED tests. The SD model was built to investigate three scenarios for the use of MCED testing: firstly, to explore the viability of MCED testing as a substitute among individuals who are not opting for SOC screening for any reason; secondly, to implement MCED testing exclusively for individuals ineligible for SOC screening, yet have high-risk characteristics; and thirdly, to employ MCED testing after SOC screening to serve as a triaging/confirmatory tool for individuals receiving inconclusive test results. The three primary scenarios were constructed by varying diagnostic accuracy and uptake rates of MCED tests. Discussion The clinical utility and outcomes of MCED testing for screening and early detection still lack comprehensive evidence. Nonetheless, this simulation model facilitates a thorough analysis of MCED tests within the Australian healthcare context, providing insights into potential additional detections and costs to the healthcare system, which may help prioritise future evidence development. The adaptable yet novel SD model presented herein is anticipated to be of considerable interest to industry, policymakers, consumers and clinicians involved in informing clinical and economic decisions regarding integrating MCED tests as cancer screening and early detection tools. The expected results of applying this SD model will determine whether using MCED testing in conjunction with SOC screening offers any potential benefits, possibly guiding policy decisions and clinical practices towards the adoption of MCED tests.

There has been growing attention to multi-class classification problems, particularly those challenges of imbalanced class distributions. To address these challenges, various strategies, including data-level re-sampling treatment and ensemble methods, have been introduced to bolster the performance of predictive models and Artificial Intelligence (AI) algorithms in scenarios where excessive level of imbalance is present. While most research and algorithm development have been focused on binary classification problems, in health informatics there is an increased interest in the field to address the problem of multi-class classification in imbalanced datasets. Multi-class imbalance problems bring forth more complex challenges, as a delicate approach is required to generate synthetic data and simultaneously maintain the relationship between the multiple classes. The aim of this review paper is to examine over-sampling methods tailored for medical and other datasets with multi-class imbalance. Out of 2,076 peer-reviewed papers identified through searches, 197 eligible papers were chosen and thoroughly reviewed for inclusion, narrowing to 37 studies being selected for in-depth analysis. These studies are categorised into four categories: metric, adaptive, structure-based, and hybrid approaches. The most significant finding is the emerging trend toward hybrid resampling methods that combine the strengths of various techniques to effectively address the problem of imbalanced data. This paper provides an extensive analysis of each selected study, discusses their findings, and outlines directions for future research.