ISSN print edition: 0366-6352 ISSN electronic edition: 1336-9075 Registr. No.: MK SR 9/7 Published monthly

Maximizing essential oil recovery from Aframomum. citratum via spontaneous solid-state fermentation: a comparative study of RSM and ANN approaches

Sylvie Kwanga Nguikwie, Steve Olugu Voundi, Doriane Djuffo Tegoundio, Brice Thaddee Mbalale Mbalale, Victor Moussango Davy, Samuel Patrick Malle Moukouri, Olivier Choisi Mayouck Etane, Alex Brandown Wambo Talla, Marlyse Leng, Achille Bissoue Nouga, and Frederic Marie Tavea

Laboratory of Biochemistry of Natural Substances (LBNS), Faculty of Sciences, University of Douala, Douala, Cameroon

E-mail: nguikwie_sylvie@yahoo.fr

Received: 21 January 2026  Accepted: 19 April 2026

Abstract:

Aframomum citratum, a little-studied aromatic plant from Cameroon, produces essential oils rich in monoterpenes, yet their industrial use remains limited by low extraction yields. This study investigated spontaneous solid-state fermentation (SSSF) as a pretreatment to enhance essential oil recovery, combined with response surface methodology (RSM via Box-Behnken Design, BBD) and artificial neural networks (ANNs) to model and optimize the extraction yield. A 15-run experimental design was employed to evaluate the effects of fermentation time, moisture content, and hydrodistillation time. The ANN model (3-5-1 architecture using tansig and purelin transfer functions) showed slightly higher predictive performance than the quadratic RSM model (R² = 0.98 vs 0.97), reflecting its ability to capture nonlinear relationships among process variables. Model robustness was further evaluated using leave-one-out cross-validation (LOOCV), which confirmed stable predictive behavior for both approaches. Optimization using the RSM desirability function predicted a maximum oil yield of 1.73%, whereas ANN-based numerical optimization using the fmincon algorithm predicted a slightly higher yield of 1.77% at 9.89 days fermentation, 42.73% moisture content, and 46.24 min hydrodistillation. These results demonstrate that SSSF-assisted extraction coupled with data-driven modelling provides an effective strategy for improving essential oil recovery from A. citratum and supports its potential valorization.

Keywords: Aframomum citratum; Essential oil; Solid-state fermentation; Optimization; Response surface methodology; Artificial neural networks

Full paper is available at www.springerlink.com.

DOI: 10.1007/s11696-026-04931-1

Chemical Papers 80 (8) 8977–8997 (2026)