Sepsis is a life-threatening condition that affects 1.7 million adults in the United States each year and claims the lives of over 270,000 individuals. It is one of the most costly diseases in the US, with total expenditure amounting to $24 billion each year. Sepsis can be difficult to diagnose because the manifestation of symptoms varies from patient to patient and many times the early signs are similar to other common conditions in the intensive care unit (ICU). In this work, the 2019 PhysioNet Challenge datasets were used to develop a machine learning neural network in MATLAB with the goal of predicting sepsis six hours before current clinical detection methods. These datasets were used for training the model and consisted of files that contain hourly parameter measurements for over 5,000 unique patients. The process for implementing this model includes data input, cleaning, handling of missing values, and training. Time intervals around sepsis onset were selected to provide training data. The output is a probability of sepsis with a binary positive or negative prediction. For the challenge, the prediction will be scored using a utility function provided by PhysioNet. Our preliminary results show that training an artificial neural network (ANN) results in 98.6% classification accuracy testing with time interval trimmed data. The ANN used had a single hidden layer of 10 neurons. A 15/15/70 (test/validation/training) split was applied to the data to tune the algorithm. A utility score of 0.038955 was achieved by running the ANN on the original data of 5000 patients. The classification accuracy was 98.6%, which suggests that the error was not in the model but rather how the model was trained (ie. preprocessing of data, handling of missing data, time intervals used, etc.).