The use of technological resources for the development of critical tasks in the hospital environment, such as surgery rooms, must be done carefully, avoiding the contamination of materials by touch when using keyboards or mouse-controlled equipment, for example. In this sense, the use of devices that can be controlled by hand gestures appears as an appropriate approach to overcome this problem. Despite the obvious benefits, this type of interaction brings some challenges, such as the need for a vocabulary of gestures suitable for carrying out tasks and, in addition, a vocabulary of gestures that can be recognized by the sensor present in the environment. In this work, we describe the results obtained addressing the gesture vocabulary recognition task using the Leap Motion sensor, aiming to couple it in the near future to the system used in the urgency and emergency unit of the Regional University Hospital of Maringá. For this purpose, we defined a hand gesture vocabulary and a set of features composed of the distances of fingertips to the palm center of the hand. Following this, we created a hand gesture dataset, composed of 10 different hand gestures, with a total of 20,000 samples. The database created will also be available as a contribution to this work. For the classification, we evaluated many different classifiers. Experiments have shown that promising results can be achieved using the proposed strategy: through hyperparameter optimization using Bayesian search and combining models with a voting classifier, we achieved an accuracy of 95.8.