AIMS: We aimed to develop and validate classifier models to assess disease activity in UC patients by evaluating potential classifier genes expression levels. METHODS: The study comprised UC and healthy control participants undergoing colonoscopy. We screened candidate genes (TGF-β1, CEACAM1 and CD177) using Differentially Expressed Genes. We compared candidate genes expression levels with the validated UC scores. UC patients were subsequently randomly assigned (1:1) to the discovery or validation groups. A logistic regression model integrating candidate genes expression was developed using discovery group and assessed its predictive effect in validation group. RESULTS: Three candidate genes were differentially associated with UC disease activity. TGF-β1 and CD177 were used to construct the logistic regression model. The two-transcript classifier model had an area under the receiver operating curve (AUC) of 0.938 (95 % confidence interval [CI]=0.888-0.987) in discriminating between remission and active UC and an AUC of 0.919 (0.862-0.977) in discriminating between remission-mild and moderate-severe activity in UC. CONCLUSIONS: TGF-β1 and CD177 transcript levels, measured by RT-PCR, are robust classifiers for assessing disease activity in UC patients, and the measurement of these transcript levels appears to be an effective method of monitoring condition of UC patients and predicting treatment effectiveness over time.