Scholarlink

<p>Mathematical models are finding increased use in biology and particularly in the field of cancer research. With cancer posing a significant global health challenge due to its high mortality rate and complex progression, predictive models have become essential tools in understanding disease dynamics and personalizing interventions. This study delves into the intersection of mathematical modeling and artificial intelligence (AI) techniques for survival prognosis and risk stratification. It focuses on 1-year, 2-year, 3-year, 4-year, and 5-year survivability of cancer patients under different cancer stages and treatment modalities. This paper reviews neural network models used in survivability predictions related to cancer onset, progression, and treatment outcomes, with a major focus on oral cancer. This research evaluates the impact of key design elements and training strategies. Its goal is to develop a state-of-the-art solution for complex classification problems in cancer progression prediction and survival. The study emphasizes the foundational role of mathematics in achieving these advancements. Relative to commonly reported Cox proportional hazards / Random Survival Forest baselines on comparable SEER cohorts (typically 60–65% accuracy and 65–70% F1) (Kim in Sci Rep 9 (1), 6994, 2019), our best neural configurations reach 72.64-−87.86% accuracy across 5-year to 1-year horizons, implying an absolute misclassification reduction of approximately 10–23 percentage points for horizon-specific prognosis.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Harnessing AI and mathematical modeling for oral cancer survival prediction across stages

  • 期刊论文

  • 发表日期：

    2026年3月7日

查看全文
  • Dharmaraja Selvamuthu,
  • Raghav Aggarwal

摘要

Mathematical models are finding increased use in biology and particularly in the field of cancer research. With cancer posing a significant global health challenge due to its high mortality rate and complex progression, predictive models have become essential tools in understanding disease dynamics and personalizing interventions. This study delves into the intersection of mathematical modeling and artificial intelligence (AI) techniques for survival prognosis and risk stratification. It focuses on 1-year, 2-year, 3-year, 4-year, and 5-year survivability of cancer patients under different cancer stages and treatment modalities. This paper reviews neural network models used in survivability predictions related to cancer onset, progression, and treatment outcomes, with a major focus on oral cancer. This research evaluates the impact of key design elements and training strategies. Its goal is to develop a state-of-the-art solution for complex classification problems in cancer progression prediction and survival. The study emphasizes the foundational role of mathematics in achieving these advancements. Relative to commonly reported Cox proportional hazards / Random Survival Forest baselines on comparable SEER cohorts (typically 60–65% accuracy and 65–70% F1) (Kim in Sci Rep 9 (1), 6994, 2019), our best neural configurations reach 72.64-−87.86% accuracy across 5-year to 1-year horizons, implying an absolute misclassification reduction of approximately 10–23 percentage points for horizon-specific prognosis.