Abstract: SiTEL joined forced with SMS on a BIRD funded project to develop a unique platform to author medical training simulations that will be operated by free language commands. The trainee will manage a virtual team of medical professionals that must cope with real-life medical nightmares.

When a person steps on a plane, the risk of dying in a plane crash is one in 10 million. When a person is admitted to a hospital, the risk of dying or being seriously harmed by medical error is one in 300. The SiTEL-SMS joint venture wishes to contribute to change these odds.

The $1.6 trillion U.S. healthcare sector is now mired in a deep crisis related to safety, quality, cost, and access. Each year, 10% of patients admitted to hospitals worldwide will experience some type of preventable medical injury. At least half of these critical events may be prevented if medical teams are trained more effectively.

This significant, unmet training need offers both challenges and opportunities: fertile ground for the SiTEL-SMS joint innovation. Current medical teamwork training solutions (conducted in clinical simulation centers with high fidelity mannequins) are either too costly and do not scale easily with other scenarios or, if they are online, lack interactivity and feedback (i.e. are taught using PowerPoint presentations and videos).

Interactive medical simulations offer immersion, the potential for widespread adoption, and can simulate virtual patients and environments, simulating various illnesses and environmental factors, respectively. Yet, the human factors surrounding virtual or interactive medical simulations remain an issue.

The SiTEL-SMS joint innovation offers physicians and nurses the ability to "communicate" with the synthetic characters (virtual characters controlled by artificial intelligence inside the simulation) using a free language technique. Physicians and nurses need to communicate with synthetic characters the same way they communicate with each other; using speech, not game controllers or a mouse. Where communication, non-verbal gestures, and behaviors are threshold training objectives, their solution eliminates the current Human Factors boundary between a real person and a simulated, virtual medical world. Increasingly, Natural Language Processing will improve the effectiveness of medical simulations and allow physicians and nurses to freely express themselves.

SiTEl has developed a trauma ER simulated environment while SMS created a cloud-based Natural Language Understanding (NLU) Engine that provides a natural language speech interface for interactive medical simulations.

The simulations are operated by free language commands. The trainee manages a virtual team of virtual characters acting as medical professionals who must work together to resolve a critical event. To achieve this speech-driven interaction, the NLU Engine works with SiTEL of MedStar’s Synthetic Character Engine, which translates the output of the NLU Engine into appropriate synthetic character behavior (the engine calls upon SiTEL’s library of anthropomorphically accurate movements).


Developing a platform/authoring-system for the development of medical simulations/tutorials where the trainee (a senior medical physician) copes with a medical nightmare simulated scenario and instructs the relevant medical personnel (pre-programmed avatars) by natural language commands, requests and questions.
The SMS-SiTEL joint venture has two major breakthrough innovations:

1. The trainee interfaces the training simulation by free continuous speech commands exactly as in real-life situations (or equivalent to a "multi-players" training session where all real-life medical personnel cooperate via VOIP). However, there are many training advantages for a session where the medical team is simulated by pre-programmed avatars and the trainee works alone in front of the pre-programmed training simulation.

2. The tutorial does not train the medical personnel in performing mere medical procedures but trains him/her to manage a medical team in coping with a medical nightmare.


PrintTell a friend
Jump to page content