0% found this document useful (0 votes)
907 views10 pages

Project 2 Building Chatbot For Dianogtistic Center

The healthcare industry struggles with staff scarcity which has impacted operational performance at DIRECTPyth Diagnostic Centers, with high absenteeism and attrition among lab coordinators and technicians contributing to over 88% of centers not meeting productivity benchmarks; a proposed solution is to implement a Python-based machine learning chatbot to automate customer interactions and diagnostic processes to address human resource issues currently limiting operations.

Uploaded by

Niri
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
907 views10 pages

Project 2 Building Chatbot For Dianogtistic Center

The healthcare industry struggles with staff scarcity which has impacted operational performance at DIRECTPyth Diagnostic Centers, with high absenteeism and attrition among lab coordinators and technicians contributing to over 88% of centers not meeting productivity benchmarks; a proposed solution is to implement a Python-based machine learning chatbot to automate customer interactions and diagnostic processes to address human resource issues currently limiting operations.

Uploaded by

Niri
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

Motivation behind Project 2

How Machine Learning Is Transforming Healthcare


The healthcare industry is one of the most labor-intensive industries around the world. It
requires the presence of humans to take care of people at various stages of their illnesses.
One of friends was at the AI Conclave held by Amazon in 2017 in Bangalore and was
amazed to see how an acute problem of staff scarcity, which has been plaguing the
healthcare industry in the United Kingdom, has been aptly solved by creating artificial

p
tabletop bots that would take care of elderly patients needs.

hi
The artificial tabletop bots remind elderly patients to take their pills, track their
prescriptions, and track and suggest wakeup routines. At the heart of Echo Alexa (as it is
known) is the machine learning developed by the Amazon team using its cloud

ns
infrastructure Amazon Web Services (AWS). At the heart of Alexa is the Python machine
learning code that helps it to perform tasks and learn from them through a feedback
mechanism. The wonderful part of this service is that Echo Alexa is available to a common
Python developer to use and develop their own programs and products based on Amazon’s
infrastructure.
r
te
In
n
ve
Su

AI

1 DIRECTPyth is dummy entity created for this Project


AI

p
hi
r ns
te
In
n

Table 1. Two-Year Financial Synopsis Report


ve

Financial Synopsis Fiscal Year, Fiscal Year,


($US million) 2015-2016 2016-2017

Income from diagnostics 930 1221


Su

Cost of operations 1031 1354


Average operational days for 118 106
diagnostic centers
Loss due to non-operations 101 133

2
3
Su
ve
n
AI

In
te
r ns
hi
p
Table 2. Breakup of Average Operational Days (AOD) Data
Average operational No. of centers No. of centers
days meeting AOD not meeting AOD
benchmark benchmark

India 32 234

p
Oversees 13 98

hi
Consolidated 45 332
%AGE 11.9 88

ns
The breakup of average operational days data shows that on an overall global basis,
the percentage of diagnostic centers at DIRECTPyth that were meeting the benchmark
of 200 days was only 11.9 percent and the percentage of centers not meeting the AOD

r
benchmark was 88 percent. This shows the criticality and extent of the problem at
te
DIRECTPyth. It also shows that there is acute loss of revenue happening because of the
centers not meeting the AOD benchmark. In order to deep-dive into the problem further
Ms. Abbey presented data relating to human resources at the lab diagnostic facilities for
In
the DIRECTPyth organization. This data is given in Table 3.

Table 3. Human Resources Data at DIRECTPyth Labs

Human resources at DIRECTPyth Labs Lab coordinator Lab technician Lab attendant
n

Average days present 88 97 194


ve

Average days absent 26 15 8


Attrition rate (%) 12% 9% 6%
Su

We can see from Table 3 that the highest absenteeism occurs with the lab
coordinator and lab technician roles. The lab attendant has the highest average
and lowest absenteeism. The best data has been pulled from the payroll systems of
DIRECTPyth Labs, which records the daily attendance of each human resource in
the lab. The total number of days present is taken as a benchmark value of 220
productive business days. The role of lab coordinator is one that interacts with the
customer who walks into the diagnostic lab, taking them to the various diagnostic

4
AI

p
hi
r ns
te
In
n
ve

1. What do you think is the reason for the losses at DIRECTPyth Diagnostic Centers?
Su

2. What human issues are we dealing with and how do you think they can be
addressed?
3. Do you think technology can help address the issues that you identified in Question
2?
4. Will a Python-based machine learning solution be able to address these issues at
DIRECTPyth?

5
5. What is your proposed Python-based machine learning solution?

6. What business justification do you give for your solution


that you think will be acceptable by the business leaders at DIRECTPyth?

Team@Suven would suggest that you first try to independently answer these
questions and then look at the solution given below. Please note the solution we have

p
given is one among many solutions that can be given in this business scenario. It is the

hi
business owners that ultimately decide if your solution is good and is able to address
their business problem.

r ns
te
In
Now let us try to answer the questions in our own way, giving the solution we
had proposed, which was accepted by the organization.

1. What do you think is the reason for the losses at DIRECTPyth


Diagnostic Centers?
n

The reason for the losses of DIRECTPyth Diagnostic Centers was the high
operational losses, with about 88 percent of the diagnostic centers not meeting the
ve

average operational-based benchmark. The real reason for this was attributed to high
absenteeism and attrition rate by lab coordinators in the organization. Lab coordinator
absenteeism was very high at 26 percent, and attrition rate was 12 percent, whereas for
Su

the lab technician the absenteeism was 15 percent, and the attrition rate was 9 percent.
The human resources came up with five various techniques to stop attrition and
absenteeism but with very marginal results.
2. What human issues are we dealing with and how do you think
they can be addressed?

6
AI

3. Do you think technology can help address the issues that you
identified in Question 2?
In our opinion using technology in areas where human beings fail to perform at

p
an optimal level required by a business is an apt scenario for bringing in automation,
machine learning, and artificial intelligence. The good point about technology is

hi
that it can do any given task repeatedly without getting bored or tired. It can also use
intelligence on data and be trained to perform expert tasks very easily. In the case
of DIRECTPyth Diagnostic Centers, we can use technology to replace the task of lab

ns
coordinator, putting an automated lab machine on its premises, thereby letting the
customer choose the lab test that they want to perform. The automated lab machine can
run Python-based machine learning programs that help the customer choose the best
diagnostic test for their scenario.

r
4. Will a Python-based machine learning solution be able to address
te
these issues at DIRECTPyth?

Yes, a Python-based machine learning system can be used to automate the


In
entire process in the lab right from the time the customer comes in and selects
the diagnostic package to generate the lab test to the next stage of generating the
report with its meaningful interpretations for the customer. Python-based machine
learning systems can be connected to any modern automated lab testing machine to
n

get the output from it.


5. What is your proposed Python-based machine learning solution?
ve

In this Project, We are going to give a solution that is going to be short; however, it is
going to emulate the real solution that we gave to the client in their production
environment. The Intern can build upon this solution using their own ideas in the real
world for similar situations.
Su

A word of caution to the Intern: this is a very minimalistic approach that we


have taken to create a chatbot; however, it does not limit the user from extending
its functionality. Chatbots in real life are more complicated than seen here, as in
project 1 and 2.

7
AI

Business Need: The business need of the solution is to emulate the work of a lab
coordinator and interact with the user through a chat interface. We present to you some
of the use-cases that are applicable in this case study scenario.

p
hi
Pro Tip Even before you start coding, divide your work into short use-cases and

ns
get them ratified by actual users of the system as you will be developing for them.
If you skip this step, it is likely you will be developing a product that the user may
reject, as you may have misunderstood some of the requirements.

r
te
Use-Case: As a user of an automated chatbot for diabetes
diagnostics, I should be able to know my past test data records in
order for me to check my history.
In
Use-Case: As a user of an automated chatbot for diabetes
diagnostics, I should be able to know the type of tests conducted
at the center and their purpose so that my doctor and I can decide
on the test.
n

Use-Case: As a user of an automated chatbot for diabetes


ve

diagnostics, I should be able to get the results of my tests done so


that I can get a report.

Use-Case: As a user of an automated chatbot for diabetes


diagnostics, I should be able to get a recommendation on my test
Su

results so that I am able to know the future course of action.

For DIRECTPyth Diagnostic Centers, there are more use-cases that are applicable
than those given here. However, to keep the process simple, we will be considering
only these four use-cases for our chatbot.

8
AI

p
hi
r ns
te
In
n
ve

Figure 1. Process of interaction between customer and lab coordinator


chatbots

In Figure 1 we can see that the process starts when the customer walks in and
Su

meet the chatbots in the lab coordinator, so there is no human interaction or human
supervision in this process. The kiosk has sensors that detect the presence of a human
being nearby and gives out a greeting to the customer. In the next part of the process,
if it is an existing customer, the chatbot asks the customer for a membership ID. Now
the lab coordinator chatbot has to make a decision based on the input provided if the
membership id given by the customer is found in the database; then it shows the last

9
test record result to the customer. If it is not an existing customer, then the chatbot
shows diabetes tests—many that list the entire set of tests available at the lab.

Once the customer has selected the test from the list, the lab coordinator Chat-bot
allocates a Lab Technician to perform the (selected or advised ) test and the supporting
Lab attendant then produces a report. The report is auto-mailed to the customer. The
mail id was taken by the chat-bot during the introduction phase.

p
Now that we have the process flowchart for our chat-bots defined, we can go ahead
implementing it using DialogFlow.

hi
Expected outcome :
fully implemented chat-bot with its bot link like

ns
https://bot.dialogflow.com/dummy-599e-4396-a35f-0d77497b46ef-id
to be submitted for assessment by the AI Engine.

Post that you would face 5 mins MCQ test for each project, on clearing which you

r
would be awarded your Certificate of Internship
te
In
n
ve
Su

10

You might also like