Improving platelet‐RNA‐based diagnostics: a comparative analysis of machine learning models for cancer detection and multiclass classification DOI Creative Commons
Maksym A. Jopek, Krzysztof Pastuszak,

Michał Sieczczyński

et al.

Molecular Oncology, Journal Year: 2024, Volume and Issue: 18(11), P. 2743 - 2754

Published: June 17, 2024

Liquid biopsy demonstrates excellent potential in patient management by providing a minimally invasive and cost‐effective approach to detecting monitoring cancer, even at its early stages. Due the complexity of liquid data, machine‐learning techniques are increasingly gaining attention sample analysis, especially for multidimensional data such as RNA expression profiles. Yet, there is no agreement community on which methods most effective or how process data. To circumvent this, we performed large‐scale study using various techniques. First, took closer look existing datasets filtered out some patients assert collection quality. The final included platelet samples acquired from 1397 cancer (17 types cancer) 354 asymptomatic, presumed healthy, donors. Then, assessed an array different models (e.g., feature selection transcripts) pan‐cancer detection multiclass classification. Our results show that simple logistic regression performs best, reaching 68% rate 99% specificity level, classification accuracy 79.38% when distinguishing between five types. In summary, revisiting classical models, have exceeded previously used method 5% 9.65% classification, respectively. ease further research, open‐source our code processing pipelines ( https://gitlab.com/jopekmaksym/improving‐platelet‐rna‐based‐diagnostics ), hope will serve strong baseline.

Language: Английский

Study of urine-based mRNA biomarkers for early detection of nonmuscle invasive bladder cancer (NMIBC) DOI
Omid Abazari, Maryamsadat Shahidi, Parisa Dayati

et al.

Urologic Oncology Seminars and Original Investigations, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

Language: Английский

Citations

1
Personalizing perioperative therapy in muscle-invasive bladder cancer: balancing oncologic benefit, toxicity, and the risk of overtreatment DOI Creative Commons
Geehyun Song, Whi‐An Kwon,

Eui Jung

et al.

Journal of Korean Medical Association, Journal Year: 2025, Volume and Issue: 68(4), P. 215 - 227

Published: April 10, 2025

Purpose: Muscle-invasive bladder cancer (MIBC) is an aggressive with a high recurrence risk due to micrometastases. Standard treatment, neoadjuvant cisplatin-based chemotherapy followed by radical cystectomy, not suitable for all patients, many being ineligible or experiencing recurrence, alongside significant toxicity concerns.Current Concepts: The introduction of immune checkpoint inhibitors (ICIs) into the perioperative setting —including ICI use in cisplatin-ineligible adjuvant high-risk individuals, and chemoimmunotherapy either preoperative postoperative period—has demonstrated promising clinical outcomes. Additionally, preservation strategies are currently under investigation select patients who exhibit favorable treatment responses, aiming maintain quality life without compromising oncologic Nevertheless, challenges such as overtreatment, long-term toxicity, immune-related adverse events remain significant, underscoring necessity precise patient selection.Discussion Conclusion: To personalize management MIBC, it essential develop clinically implement robust predictive biomarkers. Assessment molecular residual disease using circulating tumor DNA emerging method stratify risk, guide decisions, monitor therapeutic response real time. Future research should prioritize validation these biomarkers, refinement selection criteria strategies, evaluation novel agents antibody-drug conjugates fibroblast growth factor receptor setting. Ultimately, adopting precision oncology approach will be critical balancing efficacy achieving patient-centered

Language: Английский

Citations

1
Enhancing detection and monitoring of circulating tumor cells: Integrative approaches in liquid biopsy advances DOI Creative Commons

Thanmayi Velpula,

Buddolla Viswanath

The Journal of Liquid Biopsy, Journal Year: 2025, Volume and Issue: 8, P. 100297 - 100297

Published: April 29, 2025

Language: Английский

Citations

0
Over-expression of KRT17 and MDK genes at mRNA levels in urine-exfoliated cells is associated with early non-invasive diagnosis of non-muscle-invasive bladder cancer DOI
Parisa Dayati,

Nasser Shakhssalim,

Abdolamir Allameh

et al.

Clinical Biochemistry, Journal Year: 2024, Volume and Issue: 131-132, P. 110808 - 110808

Published: July 27, 2024

Language: Английский

Citations

1
Improving platelet‐RNA‐based diagnostics: a comparative analysis of machine learning models for cancer detection and multiclass classification DOI Creative Commons
Maksym A. Jopek, Krzysztof Pastuszak,

Michał Sieczczyński

et al.

Molecular Oncology, Journal Year: 2024, Volume and Issue: 18(11), P. 2743 - 2754

Published: June 17, 2024

Liquid biopsy demonstrates excellent potential in patient management by providing a minimally invasive and cost‐effective approach to detecting monitoring cancer, even at its early stages. Due the complexity of liquid data, machine‐learning techniques are increasingly gaining attention sample analysis, especially for multidimensional data such as RNA expression profiles. Yet, there is no agreement community on which methods most effective or how process data. To circumvent this, we performed large‐scale study using various techniques. First, took closer look existing datasets filtered out some patients assert collection quality. The final included platelet samples acquired from 1397 cancer (17 types cancer) 354 asymptomatic, presumed healthy, donors. Then, assessed an array different models (e.g., feature selection transcripts) pan‐cancer detection multiclass classification. Our results show that simple logistic regression performs best, reaching 68% rate 99% specificity level, classification accuracy 79.38% when distinguishing between five types. In summary, revisiting classical models, have exceeded previously used method 5% 9.65% classification, respectively. ease further research, open‐source our code processing pipelines ( https://gitlab.com/jopekmaksym/improving‐platelet‐rna‐based‐diagnostics ), hope will serve strong baseline.

Language: Английский

Citations

0