Natural Language Processing for Patient Selection in Phase I or II Oncology Clinical Trials DOI Creative Commons
Julie Delorme,

Valentin Charvet,

M. Wartelle

и другие.

JCO Clinical Cancer Informatics, Год журнала: 2021, Номер 5, С. 709 - 718

Опубликована: Июль 1, 2021

Early discontinuation affects more than one third of patients enrolled in early-phase oncology clinical trials. is deleterious both for the patient and study, by inflating its duration associated costs. We aimed at predicting successful screening dose-limiting toxicity period completion (SSD) from automatic analysis consultation reports.We retrieved reports included phase I and/or II trials any tumor type Gustave Roussy, France. designed a preprocessing pipeline that transformed free text into numerical vectors gathered them semantic clusters. These document-based were then fed machine learning model we trained to output binary prediction SSD status.Between September 2012 July 2020, 56,924 used build dictionary 1,858 or inclusion train (72%), validate (14%), test (14%) random forest model. Preprocessing could efficiently cluster words with proximity. On unseen cohort 264 reports, performances reached: F1 score 0.80, recall 0.81, area under curve 0.88. Using this model, have reduced screen fail rate (including period) 39.8% 12.8% (relative risk, 0.322; 95% CI, 0.209 0.498; P < .0001) within cohort. Most important clusters predictions comprised related hematologic malignancies, anatomopathologic features, laboratory imaging interpretation.Machine conservation promising tool assist physicians selecting prone achieve

Язык: Английский

Anticancer activity and QSAR study of sulfur-containing thiourea and sulfonamide derivatives DOI Creative Commons
Ratchanok Pingaew, Veda Prachayasittikul, Apilak Worachartcheewan

и другие.

Heliyon, Год журнала: 2022, Номер 8(8), С. e10067 - e10067

Опубликована: Авг. 1, 2022

Sulfur-containing compounds are considered as attractive pharmacophores for discovery of new drugs regarding their versatile properties to interact with various biological targets. Quantitative structure-activity relationship (QSAR) modeling is one well-recognized in silico tools successful drug discovery. In this work, a set 38 sulfur-containing derivatives (Types I-VI) were evaluated vitro anticancer activities against 6 cancer cell lines. findings indicated that compound 13 was the most potent cytotoxic agent toward HuCCA-1 line (IC50 = 14.47 μM). Compound 14 exhibited 3 investigated lines (i.e., HepG2, A549, and MDA-MB-231: IC50 range 1.50-16.67 10 showed best activity MOLT-3 1.20 μM) whereas 22 noted T47D 7.10 Subsequently, six QSAR models built using multiple linear regression (MLR) algorithm. All constructed provided reliable predictive performance (training sets: Rtr 0.8301-0.9636 RMSEtr 0.0666-0.2680; leave-one-out cross validation RCV 0.7628-0.9290 RMSECV 0.0926-0.3188). From modeling, chemical such mass, polarizability, electronegativity, van der Waals volume, octanol-water partition coefficient, well frequency/presence C-N, F-F, N-N bonds molecule essential key predictors compounds. summary, series promising fluoro-thiourea (10, 13, 14, 22) suggested potential molecules future development agents. Key knowledge obtained from be advantageous suggesting effective rational design related improved bioactivities properties.

Язык: Английский

Процитировано

14

Augmenting randomized clinical trial data with historical control data: Precision medicine applications DOI Creative Commons
Boris Freidlin, Edward L. Korn

JNCI Journal of the National Cancer Institute, Год журнала: 2022, Номер 115(1), С. 14 - 20

Опубликована: Сен. 26, 2022

Abstract As precision medicine becomes more precise, the sizes of molecularly targeted subpopulations become increasingly smaller. This can make it challenging to conduct randomized clinical trials therapies in a timely manner. To help with this problem small patient subpopulation, study design that is frequently proposed trial (RCT) intent augmenting RCT control arm data historical from set patients who have received treatment outside (historical data). In particular, strategies been developed compare outcomes across cohorts treated standard (control) guide use analysis; lessen potential well-known biases using controls without any randomization. Using some simple examples and completed studies, we demonstrate commentary these are unlikely be useful applications.

Язык: Английский

Процитировано

14

Design a Clinical Research Protocol: Influence of Real-World Setting DOI Open Access
Jonathan Cimino,

Claude M. J. Braun

Healthcare, Год журнала: 2023, Номер 11(16), С. 2254 - 2254

Опубликована: Авг. 10, 2023

The design of a clinical research protocol to evaluate new therapies, devices, patient quality life, and medical practices from scratch is probably one the greatest challenges for majority novice researchers. This especially true since high-quality methodology required achieve success effectiveness in academic hospital centers. review discusses concrete steps necessary guidelines needed create structure protocol. Along with methodology, some administrative (ethics, regulatory people-management barriers) possible time-saving recommendations (standardized procedures, collaborative training, centralization) are discussed.

Язык: Английский

Процитировано

7

Culture of cancer spheroids and evaluation of anti-cancer drugs in 3D-printed miniaturized continuous stirred tank reactors (mCSTRs) DOI
Salvador Gallegos-Martínez, Itzel Montserrat Lara-Mayorga, Mohamadmahdi Samandari

и другие.

Biofabrication, Год журнала: 2022, Номер 14(3), С. 035007 - 035007

Опубликована: Март 28, 2022

Abstract Cancer continues to be a leading cause of mortality in modern societies; therefore, improved and more reliable vitro cancer models are needed expedite fundamental research anti-cancer drug development. Here, we describe the use miniaturized continuous stirred tank reactor (mCSTR) first fabricate mature spheroids (i.e. derived from MCF7 cells, DU145 mix cells fibroblasts), then conduct assays under perfusion. This 3 ml mCSTR features an off-center agitation system that enables homogeneous chaotic laminar mixing at low speeds support cell aggregation. We incubated suspensions for d ultra-low-attachment plates allow formation discoid aggregates (∼600 µ m diameter). These were transferred into mCSTRs continuously fed with culture medium. characterized spheroid morphology expression relevant tumor biomarkers different maturation times up 4 weeks. The progressively increased size during 5–6 reach steady diameter between 600 800 m. In proof-of-principle experiments, demonstrated this testing. Three drugs commonly used breast treatment (doxorubicin, docetaxel, paclitaxel) probed concentrations MCF7-derived spheroids. these evaluated viability, glucose consumption, morphology, lactate dehydrogenase activity, genes associated resistance ( ABCB1 ABCC1 ) anti-apoptosis Bcl2 ). envision agitated as tumor-on-a-chip platform efficacy safety testing novel possibly personalized medicine applications.

Язык: Английский

Процитировано

11

Natural Language Processing for Patient Selection in Phase I or II Oncology Clinical Trials DOI Creative Commons
Julie Delorme,

Valentin Charvet,

M. Wartelle

и другие.

JCO Clinical Cancer Informatics, Год журнала: 2021, Номер 5, С. 709 - 718

Опубликована: Июль 1, 2021

Early discontinuation affects more than one third of patients enrolled in early-phase oncology clinical trials. is deleterious both for the patient and study, by inflating its duration associated costs. We aimed at predicting successful screening dose-limiting toxicity period completion (SSD) from automatic analysis consultation reports.We retrieved reports included phase I and/or II trials any tumor type Gustave Roussy, France. designed a preprocessing pipeline that transformed free text into numerical vectors gathered them semantic clusters. These document-based were then fed machine learning model we trained to output binary prediction SSD status.Between September 2012 July 2020, 56,924 used build dictionary 1,858 or inclusion train (72%), validate (14%), test (14%) random forest model. Preprocessing could efficiently cluster words with proximity. On unseen cohort 264 reports, performances reached: F1 score 0.80, recall 0.81, area under curve 0.88. Using this model, have reduced screen fail rate (including period) 39.8% 12.8% (relative risk, 0.322; 95% CI, 0.209 0.498; P < .0001) within cohort. Most important clusters predictions comprised related hematologic malignancies, anatomopathologic features, laboratory imaging interpretation.Machine conservation promising tool assist physicians selecting prone achieve

Язык: Английский

Процитировано

13