After Hours Doctor's Office : ETel Mashup



24 Jul 2007

I’m a little ashamed to admit it, but I was having so much fun putting this mashup together, it never occurred to me that anyone else would want to know anything about it, at least not so quick. So, I hope you’ll excuse my lack of documentation here. I’ll post the source code after the E-Tel demo.

Application Name After Hours Doctor’s Office

Contestant Thomas Howe, Thomas Howe Consulting

Resources I’ve uploaded the Powerpoint Presentation here and the recording of the IVR part of the presentation here.

What I’m trying to do This mashup had three major goals for me :
  1. To prove the assertion that integration of real time communications into the business process saves money for the business, while increasing quality of service for the customer.
  2. To explore the use of artificial artificial intelligence, by way of Amazon Turks, and it’s possible role in telephony applications
  3. To explore the interaction between modern Web 2.0 technologies and modern communications.

Business Problem : From the Healthcare intelligence network… “According to the National Center for Health Statistics, 55 percent of the 90 million visits to EDs in the United States in 1996 were unnecessary. In healthcare dollars, that means that 40.5 million people paid up to three times as much for routine care at the ED as they would have paid at a physician’s office.” According to Blue Cross, Blue Shield, an average emergency room visit costs $1,049.00, whereas the average doctor’s office visit costs $149.00. A little math shows us that we have a $44 billion dollar problem with unnecessary emergency room visits.

If a patient is sick after normal office hours, the only choices are to call the answering service to schedule an appointment for the morning, or to go to an emergency room. Many patients are unaware of which is the right choice, so they end up going to the emergency room needlessly, which not only drives up costs for the HMO, but also gives a lower quality of care for the patient and every other patient waiting in the same room. An easy-to-use, triage system to determine the proper course of care would result in faster service, lower costs and healthier patients.

Customer Experience : Mr. Kraus feels sick and dizzy, with a little left side weakness on a Tuesday night. He calls his doctor, Dr. McCarthy, to schedule an appointment for tomorrow. An IVR answers, telling the caller that it’s after hours, and asking if the call is because of routine business or because of illness. If illness, it asks if the caller is on a cell phone or not. It then asks for a voice message to be relayed to the doctor, then hangs up.

Immediately after the call is over, the patient gets an SMS message on their cell phone telling him that the call was received, and that we are forwarding the message over to the nurses. If the caller is calling from a PSTN phone, we would do an outbound call back. (I couldn’t do this one, because I’m not a real Tell Me developer, and outbound dialing is restricted for those who aren’t.) The message is sent to a bank of nurses, who listens to the message to determine if it’s urgent or not. If they think it is routine, they indicate that on their console, which results in another message being sent to the cell phone telling the patient that the matter is probably routine, and they would get a call in the morning. If urgent, the patient would get a message like “A nurse thinks you need to speak with a doctor. We are looking for one now – stay near the phone.” Urgent issues are forwarded to the doctor as an SMS message to their cell phone with a summary of the call done by the nurse. In this example, the doctor’s message would be : Mr. Kraus – 40 WM - left side weakness, nasuea -508 364 9972. The doctor would simply press the send button on his phone to call the patient.

At no time would the patient be more than a few minutes from feedback, and make the prospect of going to the emergency room so slow and painful, that they would prefer to sit and wait for the text message to get back.

The Mashup Components:
  • Tell Me VxML for the inbound calling IVR, the outbound status messages if I was a developer and to record and post the patient voice message.
  • Strike Iron Global SMS for the text messaging between patient, nurse and doctor
  • Amazon Web Services to setup the bank of nurses making determinations of urgency, and to transcribe the original voice mail by the patient for the permanent medical record.
Compelling Business Ideas and Notes :
  • Using IVR on the front end gets the critical information from the patient quickly and without asking for new patient behaviors, all without a real human doing anything.
  • Using a bank of amazon turk nurses leverages the tens of thousands of stay at home nurses with small children who wouldn’t mind making $3.00 in two minutes by listening to a voice message and determining if it’s important or not.
  • A typical doctor might have ten calls a night, which is not enough volume to pay an on-site nurse. A thousand doctors have ten thousand calls a night, which supports an amazon turk community easy.
  • An HMO would pay $300.00 to keep a non-urgent patient out of an emergency room.