OBJECTIVES: To develop and validate predictive models for estimating risk of early diagnosis of knee osteoarthritis (OA) by weighted cumulative exposure (WCE) function scores of prior knee pain consultations. METHODS: Both derivation and validation datasets were from an electronic healthcare record (EHR) database (Consultations in Primary Care Archive [CiPCA]) in England. WCE functions for modelling cumulative effect of time-varying knee pain consultations weighted by recency was derived as predictive tool in a population based case-control sample and validated in a prospective cohort sample. Two sets of WCE function scores: WCE (Half-Normal) score and WCE (Spline) score were evaluated and compared on model fitness, discrimination, and calibration both in derivation and validation phases. RESULTS: People with the most recent and the most frequent knee pain consultations were more likely to have high WCE scores (both sets) and these were associated with increased risk of knee OA diagnosis both in derivation and validation phases. Better model fit, discrimination, and calibration were observed for models with WCE (Spline).  CONCLUSIONS: WCE functions can be used to model pre-diagnostic symptoms within routine EHR data and may provide novel low-cost predictive tools that may contribute to early diagnosis.