Intelligent fluorescence detection for disease diagnosis has become a research hotspot. In the era of big data, machine learning (ML) for analyzing data and mining will be widely used in drug and biomarker detection. A novel hydrogen-bonded organic framework (HOF) HOF-DBA with good luminescence properties was successfully prepared from aromatic tetracarboxylic acid (4,4′-(anthracene-9,10-diyl)dibenzoic acid) by a solvothermal method. HOF-DBA acted as a fluorescent sensor to quantitatively identify the concentration of nitrofurazone (NFZ) by photo-induced electron transfer (PET) and competitive absorption. The detection limit was lower than 0.002 μg mL⁻¹, with high sensitivity and good reproducibility. HOF-DBA also exhibited highly efficient turn-up fluorescence sensing of γ-aminobutyric acid (GABA osteoporosis biomarker) in aqueous solution and serum. In addition, a back-propagation neural network (BPNN) model based on HOF-DBA and GABA was constructed for the first time. The actual test data showed that BPNN could accurately distinguish GABA concentrations by the maximum depth likelihood method. This work provides new insights into HOF-based sensors and combines fluorescence sensing with deep ML to achieve intelligent fluorescence detection of GABA.