From bytes to bismillah

In an era where consumer trust and food safety are paramount, the halal industry stands at a crucial crossroads. With the global halal market steadily expanding, valued at over US$2 trillion annually, ensuring the integrity of halal products is vital.

In this pursuit, data science emerges as a powerful tool, offering innovative solutions to uphold halal certification standards and combat food fraud. Data science is a multidisciplinary field that uses scientific methods, processes, algorithms, and systems to extract knowledge and insights from structured and unstructured data in vast volumes recorded as bytes – often measured in gigabytes, terabytes, or petabytes.

Data science combines various aspects of statistics, data analysis, machine learning, and related methods to understand and analyse actual phenomena with data in bytes.

The full potential of data science is being realised mainly due to the advent of big data. By providing extensive datasets that enable sophisticated analytics and insights, big data significantly enhances data science applications in the halal sector, paving the way for a more transparent and secure halal industry.

This enhancement spans several areas, including supply chain management, regulatory compliance, customer insights, predictive analytics, and product development. Integrating big data and data science promotes informed decision-making, strengthens consumer trust, supports halal product growth and diversification, and paints a promising future for the halal industry.

Allah SWT said in the Quran: “Allah is the One Who created seven heavens in layers, and likewise for the earth. The divine command descends between them so you may know that Allah is Most Capable of everything and that Allah certainly encompasses all things in His knowledge.” (Quran, At-Talaq, 65:12).

This verse emphasises Allah’s power and knowledge, as well as the intricate design of the universe. Technology such as data science reflects the vast knowledge that Allah SWT bestowed to benefit humankind. Using such technology can increase halal food compliance in the halal industry, thus positively affecting the halal industry.

This article delves into three key ways in which data science assists in this compliance:

• establishing a halal food database,
• detecting fraud patterns, and
• implementing predictive maintenance in halal facilities.

ESTABLISHING A HALAL FOOD DATABASE

Establishing a comprehensive halal food database fortified by data science principles is at the heart of ensuring halal integrity. The Malaysian Halal Management System 2020 requires the establishment of an ISO 17025-accredited laboratory before the database is established.

Upon fulfilling the accreditation requirement, the laboratory can validate and verify the analytical methods for various halal ingredients, especially the critical ones, e.g. protein and fat-based ingredients.

For products, the laboratory can work to establish analytical methods to detect alcohol in fermented food and beverages, as well as the presence of porcine in leather and hair-based products.

Various critical reviews divide the analytical methods into protein, fat and alcohol-based analytical techniques, and analysts can employ these analytical methods to analyse the ingredients and products that undergo the application for halal certification.

Then, analysts compile the results and segregate them into sources. This compilation becomes the database, which grows with the continuous addition of new ingredients and product results.

When compiling the results, all ingredients and products with porcine present can be flagged and kept in the database. These results furnish invaluable insights into the halal status of products, serving as a cornerstone for data-driven decision-making.

The database could be expanded to store genetic information, the biochemical composition of plant – and animal-based sources, and information on the presence or absence of non-halal components.

The biochemical compositions of synthetic ingredients can also be kept for reference.

The authorities related to halal could emulate the effort of establishing a food
composition database (https://myfcd.moh.gov.my/) by adding the manufacturers’ names and addresses, halal certificate from worldwide halal certification bodies, certificate of analysis, material safety data sheet and process flow of the ingredients, which are part and parcel of the raw material master list as stated in the Malaysian Halal Management System 2020.

Meanwhile, the enforcement agencies conducting surveillance audits of the halal supply chain can collect all ingredients and submit them for laboratory testing to update the database.

By integrating this disparate information, data scientists construct a holistic view of the halal supply chain, enhancing transparency and accountability.

FRAUD PATTERN DETECTION

From the established halal food database, the data scientists carry out a meticulous analysis involving leveraging advanced analytics to address and mitigate the issue of detecting fraudulent practices that could compromise the integrity of halal products.

Data scientists discern subtle anomalies within the ingredient lists and product labelling via sophisticated machine-learning algorithms trained on the established halal food database.

These algorithms can identify intricate patterns indicative of fraudulent behaviour. For instance, consider the case of a halal meat supplier whose products exhibit inconsistent labelling across different batches.

By scrutinising these discrepancies, data scientists can uncover potential food fraud cases, empowering halal certification bodies to safeguard consumer interests and ensuring the authenticity of halal products.

Likewise, halal certification bodies with access to the halal food database can employ laboratory test results to identify the possible adulteration occurrence. Discriminant analysis and machine learning approaches have been proven to assist with authentication. It can also be used to analyse halal food databases and classify the halal status of ingredients and products into their sources, i.e. plant, animal, synthetic, etc.

In the process, the database is divided into training, validation, and testing datasets. Based on the training dataset, a discriminant model is then established.

Data scientists evaluate the efficacy of the discriminant model to differentiate between the animal, plant or synthetic sources in the validation and testing datasets, where the discriminant model is improved until the ability to discriminate the sources achieves 100 per cent correct classification.

Then, this discriminant model can be used to identify and predict the potential occurrence of food fraud.

Employing a discriminant model is advantageous since it can identify the critical biomarkers for the halal authentication of each critical ingredient and product. For instance, discriminant analysis has been used to authenticate fish, bovine, and porcine gelatine sources.

The Department of Standard Malaysia has initiated the development of Malaysian Standards focusing on discriminant analysis for halal authentication.

As the Malaysian Standard is commonly referred to by the Standards and Metrology Institute for Islamic Countries (SMIIC) and upgraded into international standards, worldwide halal certification bodies and enforcement agencies could leverage discriminant data science analysis insights to curtail food fraud’s proliferation.

PREDICTIVE MAINTENANCE IN HALAL FACILITIES

The halal assurance system requires establishing the process flow of halal manufacturing and identifying and monitoring halal control points.

These activities include monitoring the operation of facilities via manual or real-time sensor readings, where production executives record the maintenance information of the machines, conveyors, storage, etc. and keep the information as an internal facility database.

Data scientists evaluate the occurrence of machinery failures in the database and train predictive models to detect subtle anomalies and deviations that may signal underlying issues concerning machinery performance.

Thus, the predictive models could accurately forecast potential failures by discerning patterns within halal facilities. They can also be implemented as predictive maintenance to guard against unexpected equipment failures and ensure uninterrupted production while upholding halal standards.

These predictions empower maintenance teams to proactively address looming issues before they escalate into costly downtime or compromise product quality.

By maintaining machinery proactively, companies mitigate the risk of production disruptions that could jeopardise halal certification and consumer trust.

Companies that adopt the incorporated predictive maintenance and data science can reduce downtime on critical production lines by up to 30 per cent (Moore, 2024), a 20 per cent decrease in maintenance costs, achieving a 15 per cent increase in asset lifespan, a 10 per cent reduction in replacement costs and a 10 per cent increase in overall production uptime (Elkateb et al., 2024; Uzoigwe, 2024).

This effort shifts from reactive to proactive maintenance strategies in the halal facilities.

TRANSFORMATIVE FORCE

While data science emerges as a transformative force fortifying the integrity of the halal industry, it is essential to acknowledge its potential challenges and limitations.

The successful implementation of data science in halal food compliance requires the willingness of the halal certification bodies to establish the halal food database, utilise the data science for halal certification, develop and maintain a robust data infrastructure, nurture skilled data scientists, and live a culture of data-driven decision-making.

Data privacy and security are crucial when handling sensitive information such as genetic data and supply chain records. Despite these challenges, the strategic integration of data science is critical to ensuring the authenticity and credibility of halal products globally.

The data science bytes shall assure one to consume the halal products with complete trust…Bismillah…

-@Halal