1

We want to ask the patient for the dental status. The patient would ask

I need an implant for the right front tooth.

We have to assign this to a tooth number. In the numbering scheme it would be tooth 11, with entity implant.

Question, should we handle the 32 teeth as intents or as entities?

The patient would not be able to give the tooth position in numbers, would take some questions to define the tooth position with the IBM Watson Assistant chatbot. As I am unsure about the answers intents would be more suited, to define an individual tooth position?

Below the schema and number for teeth. https://en.wikipedia.org/wiki/Dental_notation

             Numbering Teeth of Dentition

        upper right - 1             upper left - 2
    18 17 16 15 14 13 12 11 | 21 22 23 24 25 26 27 28 
 R --------------------------------------------------- L
    48 47 46 45 44 43 42 41 | 31 32 33 34 35 36 37 38 
        lower right - 4             lower left - 3 


 I - incisor
 C - canine
 P - premolar
 M - molar
data_henrik
  • 16,724
  • 2
  • 28
  • 49
medlibre
  • 13
  • 3

2 Answers2

1

So this sort of problem is solved easier with a custom mini-form within the chat. For example an interactive row of teeth they can click on.

Simon O'Doherty
  • 9,259
  • 3
  • 26
  • 54
  • Thanks, I forgot to add that we have an omnichannel concept, will mostly work with voice (Twilio, also Twilio SMS). We use the chat to train the questions, combined with Intercom chat. – medlibre Jan 25 '21 at 18:41
  • It would be the front end for open source image guided surgery software we develop. https://vimeo.com/245581787. Upper teeth with augmented reality visualisation, the teeth are "glued" to the glasses, you see 3 markers on the glasses. https://vimeo.com/245580172 – medlibre Jan 25 '21 at 19:15
0

See this overview about IBM Watson Assistant implementation steps for what makes an intent or entity. The intent is for the goal the user wants to achieve, entities identify objects and provide the processing context.

With that, in your scenario the intent is to identify teeth (intent), but you would do so by individual position or region (entity), right?

You could present a list of options to the user. Which region? Provide four to click or select. Which tooth? Label them for selection. This is kind of a guided selection.

data_henrik
  • 16,724
  • 2
  • 28
  • 49
  • So the anatomical region is the most important aspect. The patient might say on the phone/chat I have pain in the upper jaw. Then we try to define the region. The patient her/himself might change the location during conversation too. Because of this fuzzy situation we thought that an intent is preferable. Once defined the intent would be tooth region e.g. 11, the entity would be tooth, missing tooth, pain, inflammation and so on. – medlibre Jan 25 '21 at 18:57
  • The intent is to locate the tooth or teeths, the object (intent) is the actual tooth or region. As Simon pointed out, a form might help. I added to my answer with some more ideas. – data_henrik Jan 26 '21 at 06:37
  • Guided selection probably the best way, AI driven smart nudging. Will dive into this. – medlibre Jan 26 '21 at 11:21
  • Ok, good luck, Please remember to mark this question answered. – data_henrik Jan 26 '21 at 12:01