NaidaTania / LabCanHack-2020

NLP API to suggest symptoms and diagnonsis

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Nurse Watson - Backend

A video demo of the product can be found here
Deployed on: http://184.172.214.69:31234/

Endpoints:

/symptoms

POST:

  • Input: the user description of their health problem
{
	"uid" : "1234xxxx",
	"text" : "I have a headache",
	"suggestions" : []
}
  • Output: A list of possible medically worded symptoms
{
  "uid" : "1234xxxx",
  "suggestions": [
        [
            "Headache",
            "Headache after trauma",
            "Headache and weakness",
            "Confusion and headache"
        ],
        [
            "Neck ache or pain",
            "Shoulder ache or pain"
        ]
    ]
}

/predict

POST:

  • Input: A list of the symptoms that the user has confirmed
{
	"uid" : "1234xxxx",
	"text" : "I have a headache",
	"suggestions" : ["Headache after trauma","Neck ache or pain"]
}

  • Output: A list of other symptoms related to possible diagnoses to be recommended to user based on Apriori Algorithm
{
    "uid": "1234xxxx",
    "text": "You may experience these commonly related symptoms?",
    "suggestions": [
        "Unsteady gait (Trouble walking)",
        "Vertigo (Room spinning)",
        "Loss of balance",
        "Headache and weakness"
    ]
}

/diagnosis

  • Input: A list of the symptoms that the user has confirmed through /predict
{
	"uid" : "1234xxxx",
	"text" : "I have a headache",
	"suggestions" : ["Head injury","Headache after trauma"]
}

  • Output: A list of diagnoses and the percentage confidence based on the matching symptoms from the dataset
{
    "uid": "1234xxxx",
    "text": "Your possible diagnosis based on percentage of symptoms",
    "diagnosis": {
        "Basilar skull fracture:broken skull": "50%",
        "Cerebrospinal fluid rhinorrhea:leakage of brain fluid from the nose": "67%",
        "Concussion": "15%",
        "Epidural hematoma:bleeding around brain or spine": "17%",
        "Hemotympanum:blood behind ear drum": "50%",
        "Intracranial hemorrhage:bleeding in brain": "29%",
        "Post concussive syndrome:symptoms after a concussion": "33%",
        "Skull fracture:broken bone": "67%",
        "Subarachnoid hemorrhage:brain bleeding": "40%",
        "Subdural hematoma:bleeding on brain": "12%"
    }
}

Dataset used: https://www.kaggle.com/plarmuseau/sdsort#diagn_title.csv

About

NLP API to suggest symptoms and diagnonsis


Languages

Language:Jupyter Notebook 99.3%Language:Python 0.7%Language:Dockerfile 0.0%