Scholarlink

<p>Resistant starch (RS) is a key component of dietary fiber that offers significant health benefits, including improved gut health, glycemic control, and reduced risk of chronic diseases. Accurate prediction of RS content in foods is crucial for the development of functional foods aimed at promoting better health outcomes. Traditional methods of measuring RS are labor-intensive, time-consuming, and often require extensive analytical procedures. Machine learning (ML) techniques offer a promising alternative by utilizing large datasets of food composition, processing parameters, and digestion properties to predict RS content efficiently. This study explores the application of machine learning (ML) models to predict RS levels. The study analyzed 20 different varieties of rice (<i>Oryza sativa</i>) and 14 features including nutritional and functional traits to establish a correlation with RS. Analysis revealed peak viscosity (PV), hardness (HN), and gel consistency (GC) as the top three most important features contributing to the best-performing model's predictive power and provided insights into the key factors affecting RS content. Other viscosity-related metrics such as final viscosity (FV) and setback (SB) viscosity contributed moderately to the model, alongside total amylose content (TAC). Starches with higher PV often form stronger gel structures thus increasing GC and HN upon cooling, which can reduce enzyme access and slow digestion rates, subsequently high RS %. This study underscores the potential of ML models as a powerful tool to accelerate functional food innovation through accurate RS content prediction. Thus, integrating data-driven approaches to food science opens new avenues for creating healthier, functional food products with targeted health benefits, enhancing the development of diet-based solutions for improved human health.</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

From data to digestibility: prediction of resistant starch using machine learning for functional food development

  • 期刊论文

  • 发表日期：

    2025年2月4日

查看全文
  • Muskan Beura,
  • C. K. Mohammed Salman,
  • Sohel Rahaman,
  • Haritha Bollinedi,
  • Archana Singh,
  • Sonalika Ray,
  • Md. Yeasin,
  • Rishemjit Kaur,
  • Veda Krishnan

摘要

Resistant starch (RS) is a key component of dietary fiber that offers significant health benefits, including improved gut health, glycemic control, and reduced risk of chronic diseases. Accurate prediction of RS content in foods is crucial for the development of functional foods aimed at promoting better health outcomes. Traditional methods of measuring RS are labor-intensive, time-consuming, and often require extensive analytical procedures. Machine learning (ML) techniques offer a promising alternative by utilizing large datasets of food composition, processing parameters, and digestion properties to predict RS content efficiently. This study explores the application of machine learning (ML) models to predict RS levels. The study analyzed 20 different varieties of rice (Oryza sativa) and 14 features including nutritional and functional traits to establish a correlation with RS. Analysis revealed peak viscosity (PV), hardness (HN), and gel consistency (GC) as the top three most important features contributing to the best-performing model's predictive power and provided insights into the key factors affecting RS content. Other viscosity-related metrics such as final viscosity (FV) and setback (SB) viscosity contributed moderately to the model, alongside total amylose content (TAC). Starches with higher PV often form stronger gel structures thus increasing GC and HN upon cooling, which can reduce enzyme access and slow digestion rates, subsequently high RS %. This study underscores the potential of ML models as a powerful tool to accelerate functional food innovation through accurate RS content prediction. Thus, integrating data-driven approaches to food science opens new avenues for creating healthier, functional food products with targeted health benefits, enhancing the development of diet-based solutions for improved human health.