Chase Sebor faced the audience and spread his arms wide like a hockey goalie.

“You’re out in front of the goal and you just want the puck to hit you,” he said, eliciting laughter. 

The demonstration wasn’t just comic relief. Sebor, the co-founder of Venture Technologies in North Billerica, was making a point about medical device failure. Errors, he said, are treasures.

“You want to detect them, detect them, detect them,” he said. “You want a culture, in the engineering process, in which you celebrate errors.”

That brought him to the hockey analogy (a friend is a goalie).

“You’re trying to get your parts of the body in the way of these errors,” said Sebor, an embedded software design engineer and manager. “So you can feel it. If you don’t expose yourself to them and draw them out, you’ll never get enough of them.”

The key is to identify and address errors early, thus avoiding disaster.

But how do you do that effectively? Wednesday night, about 150 people gathered at the Bay Colony Corporate Center in Waltham for a panel discussion that aimed to answer the question.

“Why devices fail and what you need to do to prevent it from happening,” sponsored by the Medical Development Group, featured three speakers; Sebor, Paul Nickelsberg, president and CTO of Orchid Technologies in Maynard and Brian Stonecipher, a human factors engineer at Continuum in Newton.

Devices fail. That much is obvious. According to Nickelsberg’s presentation, the Food & Drug Association recalled 62 medical devices from January 2006 through May 2009. Nine of those (14.5 percent) were automated external defibrillators.

The reasons for the recall of one particular AED, which are available on the FDA’s website, included but weren’t limited to voice prompt problems, battery errors, not being able to find the shock button and problems finding the on/off button.

“One’s got to say right off, it’s a very difficult thing to design,” said Nickelsberg, whose electronic product development engineering firm’s specialties include medical product design, instrumentation design and power system design. “There’s a lot of constraints on designing the electronics of an AED. On the one hand there are a lot of failures. On the other hand they’re incredibly tough [to design].”

Obviously, Nickelsberg said, the creators of the defibrillator weren’t purposely negligent. But on the road from conception to production, something went wrong. Flaws — such as not being able to find an on/off switch — are usability issues, not mechanical failures.

“The real challenges,” he said, “aren’t just task-specific.”

Stonecipher, who since 2004 has worked on health and medical projects at Continuum, used a particular brand of insulin pump as an example. Technically it did what it was supposed to, but it was incredibly difficult to use. Its screen and buttons were too small for diabetics, who often have weakened fine motor skills and vision. A later version, which featured a bigger screen and simpler, more accessible controls, was more user friendly, and in turn, more successful.

Stonecipher also discussed legacy behaviors, which basically describe how a person acts as a result of prior life experiences. Legacy behaviors inevitably affect device testing, Stonecipher said.

“Everyone has prior experiences and they draw from those prior experiences in how they operate and work with the world,” Stonecipher said. “If you put a device out there that’s requiring a behavior counter to those legacy experiences, you’ve got a real problem.”

Then, without naming it, he described a medical device. This device treated a chronic health issue and needed to be filled with water daily. The device, however, had problems. Users were pouring water into the device’s air intake system because it looked like a clothes iron’s water spout. The visual cue was hard for people to shake.

All three panelists agreed that thorough testing and patience will help weed out problems. And sometimes, Sebor said, instead of waiting for errors to creep up, seeking out failure with free-form testing can be effective. He suggested (perhaps only half-jokingly) giving software that’s in development to your company’s grumpiest employee so he or she can try to break it. 

It goes back to Sebor’s hockey analogy. Exposing yourself to errors (or slap shots) allows you to catch and identify them, blocking them away like a goalie.

It also doesn’t hurt to expect the unexpected. As part of his presentation, Sebor showed an image of a woman riding a Segway — while pushing a baby carriage in the middle of a crosswalk. Yes, it was supposed to be funny, but he had a larger point.

Who would expect a rational human being to push her child in the street while riding a scooter? Nobody. But even the most ingenious inventions — medical devices included — can be used in unintended, unexpected ways.

“If you think your design is foolproof,” a quote in Sebor’s slideshow read, “remember: There is always a greater fool.”