Big Data: A Startup Thriller Novel

Chapter 48

When Chelsea Moore was admitted to the Stanford University Hospital at nine forty in the morning with acute myocardial infarction, nurse Helen Hope knew instinctively that she was going to need a bypass. She paged Dr. Julio Fernandez, the best cardiothoracic surgeon they had. It was unusual for a woman of Chelsea's age to have such a severe heart attack, but given her weight, it was also not a complete surprise.

Helen knew Chelsea's weight was going to cause some complications, both in surgery and recovery. But her young age should compensate for any problems. She had a good chance of surviving this. But if she didn't get herself into better shape, she probably wouldn't survive another such attack. Helen hoped Dr. Fernandez would put in a pacemaker, just in case she didn't lose the weight.

By lunchtime, Ms. Moore was out of surgery and Dr. Fernandez was working on another heart attack case. This time it was an eighty-five-year-old man with no insurance, who probably wouldn't make it.

"Ms. Moore? Can you hear me?"

Helen's voice was soft and kind. Ms. Moore opened her eyes slowly and looked around.

"Do you know where you are?"

She shook her head.

"You're at the hospital. You had a heart attack. You had triple bypass surgery, but you're okay. You're in recovery now."

Ms. Moore tried to speak, but her mouth was too dry. Helen handed her a cup of water with a tiny straw. She sucked at it and smiled.

"Take it easy. You've still got some heavy drugs in your system."

"My kids?"

"You have children? We're trying to contact—"

"No. No children. First day."

"Calm down, Ms. Moore. We'll take care of you. You're in good hands here."

Helen looked over at the monitors. The new Ancien EKG Machine Learning API had been trained on millions of example graphs just like these. It had taken decades for the technology to improve to the point that computers were better than doctors at reading these graphs. But now the accuracy was undeniable. Like Chess and Go before it, computers were now better at playing these games than the best human counterparts. Next to the charts, Ancien reported its opinion of the current patient's health and recommendations for how to proceed.

Heart Rate: Nominal

Blood Pressure: Nominal/Slightly Low

Breathing: Nominal

Oxygen Levels: Nominal

Overall Health: Nominal

Recommendation: In 2 minutes, administer 10 units IV bolus of reteplase

Helen didn't trust the machines, even though they hadn't yet given her a reason to doubt them. So far, they had been incredibly accurate. The closest thing to a mistake she'd encountered was when one of the machines recommended putting a newborn under the bilirubin therapy light. The child's skin was as clear as a peeled apple—no yellow pigment at all. She decided to ignore the recommendation. But when she checked in half an hour later, sure enough, the kid was as yellow as a lemon. She put the baby under the light and everything ended up fine, but she did wonder how she had missed it earlier.

The idea of computers trained to recognize human health problems bothered her for reasons she couldn't explain. Most of the doctors loved the machines. They trusted them almost without question. But Helen sat and watched the graphs for herself. Just in case. She was terrified of the day they would remove the charts altogether.

The graphs looked all right. She administered the recommended ten units of reteplase into the IV and went to check on the next patient.