Dr. Sheila Devasahayam is one of the authors of the newsletter paper entitled “Interpretable Machine Learning for Identifying Key Variables Influencing Gold Recovery and Grade” published in Materials (ISSN: 1996-1944).

Author’s introduction:

I am Dr. Sheila Devasahayam, a Senior Lecturer at Curtin University’s Western Australian School of Mines, specialising in minerals, energy, and chemical engineering. I hold two PhDs—one in materials science from the University of Queensland, Australia, and another in metallurgical science from the University of Madras, India. My research spans high-performance materials, polymer science, mineral processing, sustainable technologies, and the application of machine learning in resource recovery.

With over 60 peer-reviewed publications, several books and book chapters, and a patent application, I have contributed extensively to both academic and industrial advancements. My work has involved collaborations with leading organisations such as NASA, JAERI, BHP Billiton, JSPL and Moly Cop, and I have secured significant research funding for projects in critical minerals, green chemistry, and decarbonisation.

I am an active editor and reviewer for international journals, and I have served as a guest editor for Special Issues on sustainability and advanced materials. My teaching experience covers undergraduate and postgraduate courses in metallurgy, chemical engineering, and environmental policy, and I have supervised PhD and master’s students on topics ranging from mineral processing to sustainable energy.

Recognised as a Stanford–Elsevier Global Top 2% Scientist in mining & metallurgy since 2021, I am committed to advancing sustainable practices in the resources sector and fostering interdisciplinary research and education.

“Interpretable Machine Learning for Identifying Key Variables Influencing Gold Recovery and Grade”
by Sheila Devasahayam
Materials 2025, 18(18), 4318; https://doi.org/10.3390/ma18184318
Available online: https://www.mdpi.com/1996-1944/18/18/4318

The following is an interview with Dr. Sheila Devasahayam:

1. Congratulations on your published paper. Could you please briefly introduce the main research content of the published paper?
My paper explores how interpretable machine learning can help us understand which factors really drive gold recovery and grade during flotation in mineral processing. Instead of just building a predictive model, I wanted to make the process more transparent and useful for engineers. Using a small but carefully designed dataset from Ballarat gold ore flotation, I applied Gradient Boosting and SHAP analysis to highlight the most influential variables. Power, head grade, and processing time consistently stood out, and we also uncovered interesting non-linear interactions. The main message is that interpretable ML can bridge the gap between complex modelling and practical process improvement—even when data is limited.

2. What are the key takeaways you hope readers will gain from your paper?
The main takeaway is that power, head grade, and processing time are critical for flotation performance. We also found that interactions between variables, like head grade and collector type, can significantly influence outcomes. By using interpretable ML, we can move beyond “black box” predictions and provide insights that help process engineers make more informed decisions. This approach is especially valuable in mineral processing, where large datasets are often hard to come by.

3. Was there a specific experience or event in your research career that led you to focus on your current field of research?
I’ve always been fascinated by the complexity of mineral processing and the challenge of making it more efficient and sustainable. In industry, engineers often struggle with models that predict well but don’t explain why. That gap between prediction and understanding motivated me to explore interpretable machine learning, so that we can make decisions based on clear, transparent insights rather than just numbers.

4. Could you describe the difficulties and breakthrough innovations encountered in your current research?
One of the biggest challenges was working with a very small dataset, which is common in mineral processing because experiments are expensive and time-consuming. The breakthrough was showing that even with limited data, techniques like SHAP can reveal meaningful patterns and interactions. This opens the door for practical applications of AI in mineral processing without waiting for massive datasets.

5. Does technological progress provide new opportunities for the topic you are researching? Does it bring any potential risks? How do you think these factors will affect future research trends on this topic?
Technology is creating exciting opportunities—especially with explainable AI and real-time monitoring. These tools can help support process improvements and reduce energy use. But there are risks too. If we rely too heavily on models without validating them experimentally, we could make poor decisions. The future will be about combining AI with domain expertise to ensure insights are both accurate and actionable.

6. How do you evaluate research trends in this field, and what advice would you give to other young researchers?
The trend is clear: transparency and sustainability are becoming central in mineral processing research. My advice to young researchers is to embrace interdisciplinary approaches—combine data science with strong domain knowledge. Focus on solving real-world problems, and don’t be afraid to work with small datasets if that’s what the industry offers. Interpretable methods can still deliver big impact.

7. What appealed to you about the Materials journal that made you want to submit your paper? In your opinion, what can authors expect when they submit to Materials?
Materials stood out because of its strong reputation, rigorous peer review, and commitment to open access. It offers excellent visibility and a supportive editorial process, which is important when you want your work to reach both academics and industry professionals.

8. What is your experience publishing with Materials?
My experience was very positive. The reviewers provided constructive feedback that strengthened the paper, and the editorial team was responsive and professional. The promotion of the article through the newsletter was a great bonus—it helps the research reach a wider audience.

9. How do you think open access way of publishing impacts authors?
Open access is a game-changer. It makes research accessible to everyone—students, academics, and industry practitioners—without paywalls. This accelerates knowledge sharing and collaboration, and for authors, it means greater visibility and impact.