\subsection{Healthy Personalized Recipe Recommendations for Weekly Meal Planning}

This research paper presents a system designed to deliver personalized meal plans based on users'
dietary preferences and health needs. Its primary goal is to introduce a recommendation system that
utilizes both collaborative filtering and content-based filtering to suggest personalized meal plans,
making real-time adjustments based on users' dietary history and preferences. The summary, analysis,
and relationship of this research article with AI-Driven Nutrition Guide\textbf{ }are discussed in the sub-
sections below.

\subsubsection{Summary of the research item}

This paper \cite{r1} describes the SHARE system, a novel recommendation tool for healthy and
personalized meal planning. SHARE generates weekly meal plans by analyzing user input on dietary
preferences, health conditions, and food interactions. It applies CF to match users with similar
preferences and CB to account for the nutritional content of recipes. These filters help the system
recommend meals that suit a user’s specific health needs, such as low-carb or vegetarian diets. The
dynamic adaptation feature allows users to adjust their meal plans interactively, with the system learning
from their feedback to improve future recommendations. The system was tested using a dataset of
recipes and user interactions from food.com, and its efficacy was validated through experiments with
both healthy individuals and those managing chronic conditions.

\begin{figure}[h] % 'h' means place the figure here

\centering % Center the figure

\includegraphics[width=0.8\textwidth]{Figures/fig1.png} % Adjust the width as needed

\caption{Overview of the process for generating weakly meal planning..} % Caption for the figure

\label{fig:share_system} % Label for referencing the figure

\end{figure}

\subsubsection{Critical analysis of the research item}

The SHARE system shows great promise in meeting personalized dietary needs by using CF and CB
methods, which allow it to adapt in real time based on user preferences. Its ability to adjust meal plans
in real time engages users and gives them more control over their diet. The system can create complete
weekly meal plans that cater to different dietary preferences and health conditions, improving user
satisfaction. However, its reliance on collaborative filtering has some drawbacks, as it depends on having
user data from similar profiles. This can affect the accuracy of recommendations, especially for users
with unique needs or those who don't have enough history in the dataset. While SHARE does well in
creating plans for chronic conditions by examining user health data, its success relies on the accuracy of
this information; missing or incorrect health details may result in poor meal suggestions. Another
strength is its ability to adjust recommendations based on user feedback, allowing the system to learn
and improve over time, making suggestions more personal and relevant. However, the research mainly
focused on grammar and structure analysis without looking at important factors like originality, novelty,
and emotional impact. A stronger approach could include real-time health data from wearable devices to
improve dietary suggestions further, helping provide timely and accurate recommendations based on
changing health needs. Using a larger dataset could make the system more reliable and effective,
ensuring it meets the diverse needs of more users. Overall, while the SHARE system lays a strong
foundation for personalized dietary management, there are chances to improve its evaluation methods
and application range to make it even more useful and effective.

\subsubsection{Relationship to the proposed research work}

The SHARE system shares many parallels with the AI-Driven Nutrition\textbf{ }Guide , particularly in
using technology to offer personalized dietary recommendations. Both systems aim to cater to
individuals with specific health conditions, providing tailored meal plans based on personal preferences
and nutritional requirements. However, the AI-Driven Nutrition Guide takes a step further by integrating
OCR for real-time ingredient extraction from food images, offering weather-based meal suggestions, and
allowing users to choose between meal plans of varying frequency (2, 3, or 4 meals per day).
Additionally, the AI-Driven system's focus on Pakistani cultural preferences makes it more relevant to a
specific demographic, whereas SHARE adopts a more general approach to meal planning.

\subsection{Visual Recognition of Food Ingredients}

This paper reviews advancements in visual food ingredient recognition, highlighting challenges and
future directions for improving its accuracy. It focuses on applications like health monitoring and dietary
planning, linking these innovations to AI-driven nutrition guides for personalized diet recommendations.
The summary, analysis, and relationship of this research article with AI-Driven Nutrition Guide\
textbf{ }are discussed in the sub-sections below.

\subsubsection{Summary of the research item}

The progress in visual food ingredient recognition from 2010 to 2023 is thoroughly examined in the
research paper \cite{r2}. The review looks at the approaches, difficulties, and uses of computer vision
and machine learning techniques in this area. CNNs are used for feature extraction and classification;
preprocessing and picture preparation techniques are crucial; and different datasets, like Food-101 and
UEC-FOOD100, are explored. Potential applications of visual food recognition technology are explored,
including dietary assessment, nutrition analysis, and automation in food-related industries. The review
underscores the role of these systems in enhancing public health by enabling more accurate dietary
tracking and allergen detection.

\subsubsection{Critical analysis of the research item}

The systematic review effectively integrates a wide range of studies, providing a comprehensive overview
of the current state of the art in visual recognition of food ingredients. One of its main strengths is the
detailed discussion of various preprocessing, feature extraction, and classification methods, offering
valuable insights for researchers and practitioners. The journal excellently highlights the potential
applications of these technologies in dietary assessment, nutritional analysis, and automation,
emphasizing their role in improving public health and food industry efficiency. However, the analysis
could benefit from a more critical interrogation of the limitations and biases inherent in existing
datasets, many of which are generated in controlled environments and may not accurately reflect real-
world conditions. Additionally, although CNNs are underscored as the most effective approach, the
review could explore other machine learning models and hybrid approaches to provide a more complete
understanding of potential solutions. The absence of real-world validation and practical implementation
examples is another notable weakness, as it limits the applicability of the results to real scenarios.
Furthermore, the review could offer more detailed information about the computing and hardware
requirements needed to implement these advanced techniques, which is essential for researchers
planning to deploy these systems in practical applications.

\subsubsection{Relationship to the proposed research work}

The results of the systematic review have significant implications for our AI-driven nutrition guide
project. Our project follows a reviewed methodology using CNNs to extract and classify ingredients in
real-time, leveraging state-of-the-art techniques to improve accuracy. The insights into the importance of
preprocessing and robust datasets will inform our data preparation and model training processes.
Additionally, our project addresses some of the gaps identified in the review, such as integrating users’
health records for personalized food recommendations and focusing on culturally relevant meal plans
tailored to the Pakistani market. By incorporating advanced features like dynamic diet plans, a user-
friendly interface and robust data security, our project aims to provide a comprehensive and practical
solution that extends beyond the current scope of research in visual food ingredient recognition.

\subsection{Potential Impacts of Synthetic Food Dyes on Activity and Attention in Children}

This article \cite{r3} focuses on the potential effects of synthetic food dyes on children's behavior,
particularly concerning activity levels and attention spans. It reviews evidence from human clinical trials
and animal studies, highlighting a significant association between the consumption of these dyes and
increased hyperactivity in children. The authors emphasize that many of the studies indicate that the
current acceptable daily intakes (ADIs) established by the FDA may not adequately protect
neurobehavioral health. The article calls for updated research and regulatory assessments to better
safeguard children's well-being, emphasizing the importance of informed dietary choices to mitigate
risks associated with synthetic food dye consumption.

\subsubsection{Summary of the Research Item}

The paper titled \textit{Potential Impacts of Synthetic Food Dyes on Activity and Attention in Children: A
Review of the Human and Animal Evidence}, focusing particularly on children. The study was prompted
by increasing concerns about the role of synthetic dyes in exacerbating hyperactivity and attention
disorders, such as Attention Deficit Hyperactivity Disorder (ADHD). It systematically reviewed 27 human
clinical trials, 25 of which were challenge studies, alongside numerous animal studies Of the human
studies, 64\% showed evidence of a positive association between food dye exposure and adverse
behavioral outcomes, including hyperactivity and reduced attention. Many of these associations were
statistically significant. The animal studies reinforced these findings, highlighting potential neurological
effects due to dye consumption.

The review further critiques the current Acceptable Daily Intakes (ADIs) established by the FDA, arguing
that they are outdated and based on studies not designed to assess neurobehavioral impacts.
Specifically, for some dyes, the ADIs might not adequately protect children’s neurobehavioral health. The
authors call for updated evaluations and more rigorous research that considers recent evidence of
behavioral effects. They also suggest that additional research is needed to establish new ADIs that
specifically address neurobehavioral risks for sensitive populations such as children.

\subsubsection{Critical Analysis of the Research Item}

This paper offers a comprehensive and well-rounded examination of the existing research on synthetic
food dyes and their effects on children's behavior. Its strength lies in the systematic review of both
human clinical trials and animal studies, allowing for a thorough understanding of the potential risks. The
review successfully integrates findings from multiple sources, making a compelling case for the need to
reassess the regulatory limits of synthetic food dyes \cite{r3}.

However, several limitations arise. One key shortcoming is the over-reliance on challenge studies, which,
although useful in controlled environments, may not fully capture the complexity of food dye exposure
in real-world settings, where children are exposed to a mix of dyes through various foods over extended
periods. The absence of long-term, real-world studies limits the generalizability of the findings.
Moreover, many of the studies included in the review focus on children already diagnosed with
behavioral disorders, making it unclear whether the results can be generalized to children without such
conditions. Another area of improvement could be the inclusion of more culturally diverse dietary
patterns, as synthetic dye exposure may vary significantly across different regions and populations.

Additionally, while the paper acknowledges the limitations of existing datasets, it could benefit from a
deeper exploration of the mechanisms through which synthetic dyes affect behavior, particularly in
combination with other dietary factors. Although the animal studies add credibility to the findings, the
translation of these results to human health effects remains challenging due to differences in exposure
levels, metabolism, and developmental stages between humans and laboratory animals.

\subsubsection{Relationship to the Proposed Research Work}

The findings of this review have direct implications for our AI-driven nutrition guide project, which aims
to provide personalized dietary recommendations. Synthetic food dyes are commonly found in
processed foods, and their potential to negatively impact children’s behavior underscores the need for
our app to focus on ingredient transparency. By utilizing image recognition technology, our app can help
users identify foods that contain these dyes and provide real-time warnings or alternatives, empowering
users to make healthier, informed decisions.

Moreover, this research highlights gaps in the current dietary recommendations related to
neurobehavioral health, particularly in children, which our project can address by offering more
nuanced, personalized nutrition plans that take into account factors such as age, health conditions, and
dye sensitivity. For example, parents of children with ADHD may find our app particularly useful in
managing their child's diet to avoid synthetic dyes that could worsen symptoms.

In the context of our focus on culturally relevant meal plans for the Pakistani market, this research
reinforces the importance of avoiding unnecessary additives in traditional diets. The potential dangers
posed by synthetic dyes could be an educational component of our application, helping users limit
processed foods and stick to healthier, culturally appropriate options. By integrating these insights, our
project can enhance the user experience and contribute to public health by promoting cleaner eating
habits and supporting evidence-based dietary recommendations that consider neurobehavioral health
risks.

\subsection{Nutritional Approach in Inherited Metabolic Cardiac Disorders}

This section reviews research advancements in the nutritional management of inherited metabolic
cardiac disorders (IMCDs). It focuses on dietary interventions that aim to improve cardiovascular
outcomes in patients suffering from metabolic disorders, including mitochondrial diseases and lysosomal
storage diseases. These disorders often have significant cardiovascular manifestations, which are
challenging to treat. The summary, critical analysis, and relevance are presented below.

\subsubsection{Summary of the Research Item}

The article \textit{Nutritional Approach in Selected Inherited Metabolic Cardiac Disorders} explores the
use of personalized dietary interventions in managing cardiovascular complications in inherited
metabolic diseases (IMDs), specifically mitochondrial and lysosomal storage diseases \cite{r4} . The
paper focuses on dietary therapies such as the ketogenic diet (KD) and supplementation with Coenzyme
Q10, L-Carnitine, and creatine, which have shown promise in managing symptoms like cardiomyopathy
and conduction abnormalities.

It presents evidence supporting the role of tailored nutrition, such as high-protein diets or modified
Atkins diets, in improving patient outcomes, including the reduction of left ventricular hypertrophy and
improvements in heart function in patients with metabolic dysfunctions. The ketogenic diet is a
prominent focus, showing potential in addressing metabolic stress, especially in mitochondrial diseases,
although further research is recommended to support its generalized use. The paper emphasizes the
importance of continuous monitoring of dietary impact using imaging techniques like cardiac MRI and
echocardiography to assess cardiac structure and function.

\subsubsection{Critical Analysis of the Research Item}

This review comprehensively evaluates the role of nutrition in inherited metabolic cardiac disorders,
providing a detailed analysis of several dietary strategies. One key strength of the article is its in-depth
review of both mitochondrial and lysosomal storage disorders and the evidence supporting dietary
interventions like the ketogenic diet. The systematic approach of evaluating various metabolic diseases
and their corresponding dietary therapies offers significant insights into potential treatments for these
challenging disorders.

However, while the ketogenic diet is shown to have promising effects on heart function, especially in
mitochondrial disorders, the article lacks substantial real-world validation in diverse populations. The
reliance on case studies and small-scale trials limits the ability to generalize these findings to larger, more
heterogeneous groups. Additionally, the absence of randomized controlled trials (RCTs) examining long-
term outcomes of dietary interventions in IMDs is a notable gap. The paper could also benefit from more
detailed guidance on how to implement these diets in clinical settings, considering the socio-economic
and cultural factors that influence patient compliance.

Moreover, the article touches on the limitations of current diagnostic tools like echocardiography and
advocates for the increased use of cardiac MRI. However, it does not provide a clear recommendation on
which modality is superior for monitoring the impact of these nutritional interventions.
\subsubsection{Relationship to the Proposed Research Work}

The findings from this research are highly relevant to the development of the Intelligent Dietary Analysis
and Personalized Nutrition Planning Application. The proposed application leverages machine learning
and image processing to extract ingredients from food images, providing users with personalized dietary
recommendations. By integrating insights from this review, such as the focus on personalized nutrition
for managing chronic conditions, our project can optimize ingredient analysis to offer real-time, tailored
meal recommendations.

Our system extends the principles discussed in the review by incorporating dynamic health profiles and
cultural preferences, particularly for the Pakistani population, where conditions like diabetes and
cardiovascular diseases are prevalent. The AI-driven application also addresses gaps such as the real-
world applicability of nutritional recommendations by analyzing food choices in uncontrolled
environments, unlike the controlled trials highlighted in the paper. Additionally, continuous dietary
monitoring, a key area identified in the research, will be enhanced through future integration with
wearable devices that can track real-time health metrics such as blood glucose levels and blood
pressure.

By providing alternatives for unhealthy ingredients and dynamically updating dietary plans based on user
health data, the system improves user engagement and compliance. Furthermore, the system offers a
robust solution to addressing both nutrient profiling and ingredient analysis, tailored to user-specific
medical and dietary needs. Through machine learning, the application ensures the continuous
adaptation of recommendations to users’ changing health goals, bridging the gap between dietary
knowledge and personalized action to enhance overall well-being.

\subsection{Personalized Meal Recommender Systems for Nutrition Guidance}

This section reviews research on meal recommender systems, particularly those designed to provide
personalized meal recommendations based on user preferences, dietary restrictions, and nutritional
needs. The focus is on systems that aim to help users adopt healthier eating habits through personalized
suggestions. The summary, critical analysis, and its relevance to AI-driven nutrition systems are discussed
below.

\subsubsection{Summary of the Research Item}

The article \cite{r5} evaluates a personalized meal recommender system designed to help users plan
meals based on their nutritional needs, preferences, and restrictions. The system, named SousChef,
offers meal plans for multiple meals across the week using recipes rather than individual food items. The
goal is to provide culturally acceptable, nutritious, and varied meal options that align with the user’s
energy requirements. The validation study explores factors such as nutritional suitability, meal diversity,
and the incorporation of user preferences and restrictions in recommendations. The findings indicate
that SousChef successfully tailors meal plans to individual user profiles, ensuring nutritional balance
while accommodating user-defined constraints. Experiments using synthetic profiles reveal that users’
preferences and restrictions are reflected in the recommendations, and the system demonstrates strong
performance in meeting nutritional needs, with minor deviations in macronutrient distribution.

\subsubsection{Critical Analysis of the Research Item}

This research makes significant contributions to personalized meal planning systems by offering a well-
validated model that incorporates a range of user-specific data, including anthropometrics and dietary
restrictions. One strength of this study is its focus on user preferences and meal variety, ensuring that
the recommendations are both nutritionally balanced and diverse. The flexibility of the SousChef system,
which allows for the planning of various meal combinations across the week, underscores its potential as
a practical tool for personalized nutrition guidance.

However, certain limitations are noted in the study. First, the validation was conducted using synthetic
user profiles rather than real-world users, which may limit the applicability of the findings. Real-world
testing would provide a clearer understanding of the system's effectiveness in practice. Additionally,
while the system performed well in terms of energy distribution, some deviations were noted in
macronutrient balance, particularly in fat content, suggesting that further optimization of recipes and
ingredient databases may be required. The reliance on the Portuguese food composition database may
also pose challenges when extending the system to other cultural or geographic contexts. Finally, the
article could have offered more detail on how the system might handle long-term dietary
recommendations or specific health conditions beyond general nutritional guidance.

\subsubsection{Relationship to the Proposed Research Work}

The findings from this study are highly relevant to the development of the Intelligent Dietary Analysis
and Personalized Nutrition Planning Application. Similar to the SousChef system, our application aims to
personalize meal recommendations by considering user preferences, nutritional needs, and dietary
restrictions. However, our project goes further by incorporating advanced image processing and machine
learning techniques to extract and analyze ingredients directly from food images, thereby enhancing
real-time meal recommendations. This feature simplifies ingredient identification while providing
tailored dietary plans based on individual health profiles and cultural preferences.

In addition, our application focuses on dynamically adapting to changes in user health data, integrating
medical conditions such as diabetes and cardiovascular diseases, which are prevalent in the Pakistani
population. By bridging the gap between dietary recommendations and real-world applicability, our
system addresses limitations identified in the SousChef study, including expanding the food composition
database to cover a wider range of cultural and regional diets. Furthermore, the proposed system will
offer alternative suggestions for unhealthy ingredients, providing users with healthier options aligned
with their medical conditions.

To further advance the capabilities of personalized nutrition systems, our project envisions future
integration with wearable health devices to monitor real-time user metrics such as blood glucose levels
and physical activity. This level of dynamic feedback and health monitoring will ensure that dietary
recommendations evolve continuously based on the user’s current health status, offering a truly
personalized and responsive solution to dietary management.

\subsection{Challenges in Personalized Nutrition and Health}

The article \cite{r6} talks about creating personalized diet plans based on individual factors. It uses
advanced technology to build models that can predict the best diet for each person. The article discusses
the challenges in doing this, such as limited data and the need for better technology. It highlights the
potential of using different data sources to create more personalized nutrition plans. However, it also
emphasizes the importance of addressing privacy concerns and ensuring the accuracy of predictions.

\subsubsection{Summary of the Researched Item}

The growing significance of personalized nutrition—which adjusts dietary recommendations based on

individual biological responses to food—is covered in the article. Numerous factors, including genetics,

metabolism, microbiome, and lifestyle, affect this personalization. A number of issues still exist in omics
technologies and machine learning, such as data sparsity, missing data, and the requirement for more
robust computational infrastructures to manage health data. The article argues in favor of

comprehensive, data-driven models and draws attention to the shortcomings of reductionist

methodologies. These models would predict health outcomes and suggest customized diets by
integrating multiple data streams (e.g. wearable, electronic health records). The objective is to enhance

general health and disease prevention by means of targeted dietary interventions.

\subsubsection{Critical Analysis of the Researched Item}

Key obstacles to personalized nutrition are effectively identified in the article, including the limitations of
reductionist methods and the requirement for computational tools that incorporate health data. Its
emphasis on using cutting-edge technologies for dietary recommendations, such as AI and ML, is in line
with contemporary developments in precision medicine. The

paper, however, should have included more useful suggestions or case studies showing these

technologies used successfully. The importance of infrastructure is often overstated, but data privacy
ethics are just as important and should receive equal attention when working with individualized health

data. The paper also points out gaps in platform-to-platform data standardization, but it doesn't go into
great detail about how to reconcile these discrepancies. Its innovative interdisciplinary approach, which
combines ideas from genomics, data science, and nutrition, is one of its strong points. Though the article
offers big data and machine learning as solutions, it downplays the difficulties that may arise when
integrating these technologies with the current healthcare systems, especially in low-resource settings

where issues with data quality and availability may arise.

\subsubsection{Relationship to the Proposed Research}

The difficulties and solutions outlined in the article are closely aligned with the research project
proposed. Both highlight how machine learning is used to provide individualized dietary

recommendations based on a person's health profile and medical records. The article calls for
personalized, localized health solutions, and the AI-Driven Nutrition Guide responds to the specific needs
of the Pakistani population by taking into account cultural and medical dietary challenges, such as

diabetes and cardiovascular diseases.

Furthermore, the emphasis on real-time ingredient extraction via image processing and the
incorporation of medical data in the proposed application aligns with the article's emphasis on the
necessity of advanced computational models to manage customized nutrition. The article's suggestion
for predictive and adaptive models that react to individual variability in nutrition and health is echoed by
the project's aim of developing a personalized diet plan that dynamically adapts to changes in health.

\subsection{Personalized Nutrition Recommendation for Diabetic Patients Using Optimization

Techniques}

The article \cite{r7} talks about using special algorithms to create personalized diet plans for diabetic
patients. It groups patients based on their health data and creates customized diet suggestions. The
results show that one algorithm is better than traditional methods at providing accurate and effective
recommendations. This study suggests that these algorithms can be helpful in healthcare, especially for
managing chronic diseases through personalized nutrition.

\subsubsection{Summary of the Researched Item}

An approach that employs optimization techniques like Ant Colony Optimization (ACO) and Particle
Swarm Optimization (PSO) to enhance personalized nutrition recommendations for diabetic patients is
presented in this research article. The study clusters patient profiles according to their health
information in order to study web-based recommender systems for nutrition therapy. In order to ensure
that patients receive individualized dietary advice, the system uses optimization techniques to analyze
the profiles and personalize recommendations. According to the experiment results, PSO performs
better than other methods, especially when it comes to accuracy, F1-measure, and Matthews
correlation.

The authors created distinct profiles based on blood glucose levels, insulin dosages, and other health
metrics using the medical records of diabetic patients. The optimization algorithms grouped patients
with comparable profiles and suggested effective nutrition therapies based on profile matching. The
system seeks to improve the efficacy of dietary management for patients with diabetes by concentrating
on increasing recommendation accuracy and optimizing patient outcomes.

\subsubsection{Critical Analysis of the Researched Item}

The complexity of individualized nutrition for diabetic patients—a critical task in the management of

chronic diseases—is skillfully discussed in this article. The innovative use of ACO and PSO algorithms is
demonstrated by the improvement in recommendation accuracy and search time reduction. However,
the study might be strengthened by adding larger datasets and real-time patient feedback, which might
increase the system's adaptability to a range of patient populations. The emphasis on a small dataset (70
patients) that might not be sufficiently representative to allow for the generalization of the results is
another drawback. Although the optimization techniques yielded good results, accuracy could be further
improved by incorporating hybrid methods or other machine learning models

\subsubsection{Relationship to the Proposed Research}

With a focus on Pakistani cuisine and chronic illnesses like diabetes, the proposed project, this article
seeks to provide real-time dietary recommendations based on cultural preferences and health
profiles.The project's and the article's common objective is to help patients with medical conditions
adopt better eating habits by focusing on personalized nutrition management. By offering insights into
the application of ACO and PSO for recommendation systems, which could be modified into the AI-
driven nutrition guide, the research described in the article could directly inform the project.
Furthermore, the method of grouping user profiles and making dietary recommendations based on
health information may improve the customized meal plans in the suggested system. Just as similar
optimization techniques improved the diabetic nutrition recommendation system's accuracy, they would
also improve the performance and accuracy of the personalized recommendations in the Pakistani
context.

\subsection{Digital Food Sensing and Ingredient Analysis Techniques to Facilitate Human-Food Interface
Designs}

The article \cite{r8} talks about different ways to analyze food. These ways use things like cameras, light,
sound, and chemicals to learn about the inside of food. The goal is to make it easy to use these
technologies in everyday life to know more about the food we eat. The article talks about the potential
of these technologies in areas like tracking what we eat and checking the quality of food. It also
discusses the difficulties in learning about the inside of food. By comparing the strengths and
weaknesses of each technology, the article suggests ways to improve food analysis in the future.

\subsubsection{Summary of the Researched Item}

The development of interactive technologies for Human-Food Interaction (HFI) is the main topic of
thisarticle. It examines digital food sensing methods and divides them into four categories:
electrochemical, spectroscopy, acoustic, and imaging. These techniques seek to examine the contents,
volume, and internal elements of food, such as the amount of sugar or salt. The main objective is to
surmount the obstacles in determining the intrinsic qualities of food and enhance the smoothness and
intuitiveness of human-food interactions. Ten sensing techniques are reviewed critically in this paper,
with special emphasis on their potential for real-time, non-disruptive food analysis while food is being
consumed. In order to improve the integration of these sensing technologies into regular dining
experiences, the authors also suggest future research directions

\subsubsection{Critical Analysis of the Researched Item}

The study thoroughly examines the state-of-the-art food sensing technologies, pointing out their
advantages and disadvantages in terms of human-food interactions. It draws attention to important

issues, such as how challenging it is to precisely sense internal food properties like ingredient
composition using current techniques like computer vision. The paper's strength is its thorough analysis
of various methods (including computer vision, electronic noses, and infrared spectroscopy) with a focus

on how they can be used in real-time dining. The essay, however, spends more time discussing technical
details than it does discussing potential difficulties with user-centric design when incorporating these
technologies into daily life. An additional constraint that restricts the practical applicability of these
sensing methods is the absence of comprehensive case studies or experimental outcomes regarding
their effectiveness in real-world situations.

\subsubsection{Relationship to the Proposed Research}

The proposal suggested research closely resembles the goals covered in the article. Both projects use
image processing and machine earning technologies to analyze food ingredients and provide
individualized dietary insights. The article's vision of using digital food sensing for dietary assessment is
directly aligned with the AI-Driven Nutrition Guide, which focuses on using food image processing to
extract ingredients and provide personalized nutrition plans, especially for people with specific health
needs like diabetes or cardiovascular diseases.

Although the article provides a wide overview of possible sensing technologies for ingredient analysis,
the research that is being suggested takes a more focused, culturally appropriate approach by

concentrating on Pakistani food and health issues. Although the article highlights the potential of non-
invasive food sensing techniques like spectroscopy and electrochemical methods, the AI-Driven Nutrition

Guide is more focused on machine learning. Despite this, both works have similar goals.
Therefore,incorporating the food sensing methods that the article suggests—like electronic tongues or
hyperspectral imaging—could improve the project's proposed ingredient analysis's precision and
accuracy, enabling more accurate and real-time dietary recommendations

\subsection{Application of machine learning for estimating label nutrients using USDA Global Branded
Food Products Database, (BFPD)}

The use of machine learning (ML) models to forecast food product nutrient values based on ingredient
lists is examined in this article \cite{r9}. The study evaluates five machine learning (ML) models:
Adaboost, Bayesian Ridge, Support Vector Machine (SVM), Multi-Layer Perceptron (MLP), and a
specialized MLP variant (MLPcr). It uses data from the USDA Global Branded Food Products Database
(BFPD). The most accurate predictions were made by MLP and MLPcr of these, especially for sodium,
protein, and carbs. The study shows how neural networks can improve the prediction of food nutrients,
laying the groundwork for tools that cater to individual dietary needs. It also draws attention to issues
with nutrient distributions and ingredient proportions, suggesting opportunities for future database
quality and model training improvements.

\begin{figure}[h] % 'h' means place the figure here

\centering % Center the figure

\includegraphics[width=0.7\textwidth]{Figures/graphical_abstract.png} % Adjust the width as needed

\caption{Graphical representation of processes performed in article} % Caption for the figure

\label{fig:share_system} % Label for referencing the figure

\end{figure}

\subsubsection{Summary of the Researched Item}

The use of machine learning (ML) models to predict food nutrient content based on ingredients is
explored in the paper. The USDA Global Branded Food Products Database (BFPD), which offers extensive
information on food ingredients and nutrition labels, is utilised in this study. Five machine learning
models were tested by the authors using a dataset they had prepared from this database: Adaboost,
Bayesian Ridge, Support Vector Machine (SVM), Multi-Layer Perceptron (MLP), and an MLP variant called
MLPcr. Predicting 13 nutrient values for food items was the aim, with an emphasis on sodium, protein,
and carbs. The most accurate results were obtained by MLP and MLPcr, with MLPcr increasing accuracy
for carbohydrates by 252 percent.The study shows that neural network models can accurately predict
nutrient values from ingredients, but it also points out that inaccurate predictions can be made due to
skewed nutrient distributions and incomplete data on ingredient proportions.

\subsubsection{Critical Analysis of the Researched Item}

The findings of this study demonstrate how machine learning, more especially neural networks, can be
used to predict food nutrients from ingredient lists alone, which is a major contribution to the field of
nutritional analysis. It emphasizes how these methods could be applied to tasks such as customized
dietary advice. The study's creative use of the USDA database and its comparison of accuracy using
several machine learning models are its strongest points. The neural network models' high accuracy,
particularly for carbohydrates, highlights the potential of AI-driven solutions in the field of food science.

But there are also restrictions on the study. First, the model's capacity to predict certain nutrients with
high accuracy is limited by the lack of proportion data for ingredients, particularly for foods with intricate
formulations like condiments. Furthermore, the study only looks at a small number of nutrients, and
because of the skewed distribution of the data, sodium prediction is less accurate. This method is
promising, but it's still in its early stages of development, as the study notes that databases and training
data need to be improved.

\subsubsection{Relationship to the Proposed Research}

The proposed research closely matches the goals of the article. The goal of both projects is to improve
nutrition planning and dietary analysis by utilizing machine learning. In order to analyze food ingredients
and provide customized dietary recommendations based on individual health profiles and cultural
preferences, the proposed research work focuses on developing an Intelligent Dietary Analysis and
Personalized Nutrition Planning Application that integrates image processing and machine learning.

The paper. can offer fundamental understandings for the project that is being suggested, especially
concerning the application of machine learning models such as MLP and MLPcr for nutrient content
prediction. Similar to the nutrient analysis problem, the AI-Driven Nutrition Guide extends the problem
to real-time ingredient extraction through image processing. The techniques covered in the study can be
useful for this extra layer of real-time interaction, particularly when predicting nutrient content from
ingredient lists.

Overall, the research work provides a methodological framework and validation for predicting food
nutrients using machine learning, which the proposed project could modify and improve to meet the
unique dietary needs and health circumstances of the Pakistani populace. The suggested work, on the
other hand, presents a more thorough, individualized dietary management system by combining medical
histories and real-time data processing.