Producers now have many technologies available for monitoring daily changes in milk composition and cow behavior to assist in disease detection. This study aimed to develop algorithms for identifying gram-negative (GN) and gram-positive (GP) mastitis using a combination of milk and activity measures. Milk yield, quality (electrical conductivity and SCC) and composition (lactose, protein, and fat percentage) were collected using an in-line milk analyzer (AfiLab, S.A.E. Afikim, Israel) at Virginia Tech (VT) and University of Florida (UF) Dairy for 14 d before and following a clinical mastitis (CM) event (n = 268). Activity measures included daily steps at UF (Afi Pedometer), as well as number of rest bouts, total resting time and rest bout duration at VT (Afi PedometerPlus). A quarter milk sample was collected for bacteriology upon detection of CM. Data were also retained for control animals matched to each clinical case (n = 268) based on breed, lactation number and DIM. Rather than using the absolute value of variables as the primary explanatory variable, slopes of each variable were estimated using linear regression over the days before CM detection. Slopes were calculated between d 7 and 5, 4, 3, 2 or 1 before infection to better understand how early these metrics could be used to detect CM. Infection was treated as a binomial response and backward stepwise elimination mixed-effect regression was used to relate infection to explanatory variable slopes. Farm was included as a random effect. Explanatory variable slopes ranging between 7 and 2 d before CM had the highest detection accuracy rate for both pathogen types. Instance of GN infections was correctly identified in the algorithm in 85% of cases. This detection algorithm included slopes of milk yield (P = 0.02), SCC (P = 0.01) and daily steps (P = 0.11). For GP infections, the most accurate model included slopes of conductivity (P = 0.07), protein percentage (P = 0.02) and SCC (P = 0.01), and correctly identified 75% of infections. The algorithms suggest activity and milk data may be potentially useful for early detection of CM.