Healthcare Decoded – The #Analytics Conundrum by Harish Rijhwani, @Harish_Rijhwani

If we want to start using Analytics in India, one of the areas to focus on can be in the area of Diagnostic Analytics. We can leverage Transfer learning in this area as there are many pre-trained models leveraged by others and available. 


Abacus, one of the earliest form of Calculators, originated around 5000 years back. If we want to look at something, somewhat near to our timeline, we need to go to around 1890. In this year the Hollerith Tabulator was created, it is an Electromagnetic machine use to summarize information present on punch cards. Fifty years from here (the 1940s), Alan Turing created the Turing machine, first of its kind Analytics model to decode encrypted German Messages. From here on the world saw a growing need for computers. Fast forward to today’s day and age; we have reached the point where people have experimented with using DNA as a storage medium. DNA - Deoxyribonucleic Acid which stores the genetic instruction of the body. Oh, by the way, a DNA can store around 215 petabytes of information which is equivalent to 215 million gigabytes. Phew! That’s a huge number. 

Over the past 120 years, Healthcare is one of the most unchartered territories when we consider Analytics. Organizations have focused on Supply Chain, Finance, Human Resources, and other domains. Compared to all these industries, Healthcare is in a stage of discovery. At some point in everyone’s life, one has been to a hospital. In my case I had gone to the US for a business trip in Feb 2006, and what do you know, I got chickenpox. I had to visit an “Urgent Care” where I waited around 20-30 minutes in the lobby before my turn came. Once my turn came, we went to the examination room where the nurse came and asked me some more questions. The nurse also had a form with her clipped on a writing pad. She asked me various questions around my lifestyle for example
  • Do You Smoke?
  • Do You Drink?
  • Are you on any medication?

Along with this, she also took my vitals in terms of Blood pressure, height, weight, and temperature. She noted down all this information on a paper form. After all of this is when arrived the Physician. The physician looked at the information filled on the chart by the Nurse, then looked at the spots and said – "You have Chicken Pox." The Physician then gave me my prescription and also gave me a date for my next visit, which was around 4-5 days away. It was now time to leave, and I had to pay a bill of approx. $200/- out of my pocket, note that this did not include the bill for medications which I had to buy from a nearby pharmacy.

Looking Back

Let us look back at some of the important aspects of the scenario depicted. If it came to your attention, the nurse was using a “Paper Chart,” this is the biggest challenge for the field of Business Intelligence and Analytics. Data in an electronic format is very much necessary for anyone to perform Analytics or create a simple Report depicting Insights viz. “No of Patients visiting the Hospital Daily.” 

The second aspect, I had to wait for 20-30 minutes before the physician could see me. The scenario depicted was not an Emergency, but wait times are important and along with it the Patient Volume, which could be Monthly/Weekly/Daily/Hourly. One of the Largest US Based IDN’s (HCA hospitals) takes the help of large Electronic Billboards to keep people informed of the wait times.

The third aspect is the final bill amount which in this case, I had to pay out of pocket. In my case, I was on a Business Trip and had Insurance but did not have an Insurance Card. In other words, you can say it was not cashless insurance. In general, it would be important to know the approximate cost of treatment upfront.

One final aspect which does not come out that is unlike any other industry; the Healthcare Industry has multiple entities viz.

  • Healthcare Provider – The Hospital / Care Provider, which can be a large hospital to a clinic.
  • Health Insurance Company 
  • Pharmaceutical – Where the patients buy medications
  • Pharma and Life Sciences – These organizations focus on developing new medications
  • Medical Devices – These organizations develop Biomedical Devices viz. Glucometer

Deep Dive

Now that you have a basic idea about Healthcare and some of the touchpoints let us deep dive into more details around some possible scenarios of Analytics which can be done/achieved. These scenarios are not limited to just a Hospital/Care provider but would cover various entities. Let us divide the scenarios across the different types of Machine Learning Algorithms and see what we could do with each specific example.

a. Regression

Regression in its simplest format can solve very limited problems but is the basis for various other algorithms. In the case of healthcare, one can use Regression for some of the below examples.

Predicting Length of Stay(LOS): The Revenue for a hospital is directly proportional to the total number of patient visits/discharges. In 2018 my mother had to be admitted for treatment of Pneumonia, and we were not sure how long would it take to get a discharge. Discharge is one aspect, but there is a certain amount of time my mother had to spend in the ICU. The uncertainty around the duration causes a lot of stress, and this is not limited to me but impacts any individual. Regression can be used to calculate the Length of Stay, which can be beneficial to the hospital for planning and scheduling but also the Patients relatives. This length of stay can be broken down by predicting the number of days in ICU as well as in a ward/room before discharge. Assuming this is a supervised learning model, some of the data elements which would be needed to effectively predict/build an algorithm would be Primary Diagnosis, Secondary Diagnosis, Comorbid Conditions, Admit Date, Discharge Date, Days in ICU, Severity of Illness, Prior Admission Details, Medications, Physician Specialty, Date of Birth, Lab Results. 

Reduce Patient Readmissions: Once we have predicted the length of stay, the next logical step would be to find out the possibility of the patient getting readmitted. Readmission is another key concern, and in countries like the US, if a patient gets readmitted within 30 days, there is a possibility of a penalty to the Hospital. The readmission could be due to various reasons viz. “Patient Refused to take therapy” or “Hospital Acquired Infection” to name a few. In this case, we would require similar information/data elements like for LOS. Also, one can consider the geographical location of the patient (Address/Zip Code). We could also leverage external databases related to the weather. E.g., Cases of Pneumonia increase when the weather suddenly changes from Hot and cold and vice versa.

Healthcare Cost Prediction: The Healthcare Cost referred to, can be the Total cost of treatment based on the diagnosis/treatment. By the way, when we talk about regression, we refer to predicting a numerical value, and in this case, one can also find the possible Insurance Premium a member would pay. In this case, as well, one would require data elements used in LOS and Readmissions. Along with this, one could also require Hospital Address/Zip Code, the Cost of Treatment (Prior Years).

Chronic Disease Prediction: At the beginning of the paper, we spoke about the Rise of Computers. In this day and age one cannot live without a Mobile phone (which is a powerful computer in itself). It was 2009, and I had gone to visit an Orthopedic Physician as I had got a stiff neck. The doctor came, asked me mover your neck to the right, left, up down and laterally. One question he asked me, “What is your Profession?” I said I work in the IT Industry (Software). To that, he promptly mentioned, “Your profession is your problem.” For anyone who works in the IT industry, there is a very high probability that a person will get some neck or back problem. By the way, when we look at the young population, kids today use mobiles from the age of 5 & 6 and the probability of getting a similar problem will be pretty high. One can use Logistic Regression to predict the probability of such cases. The data required for such cases might not be so easy to gather since we require Lifestyle details. Things like smoking and drinking are captured during the Patient visit, but what one eats daily is not captured in any system unless the patient is using some system and many such patients are willing to share this data. have 

b. Classification and Clustering

The Banking Industry uses Analytics, especially in terms of Identifying Fraud. I will give a simple example which helps with the explanation. I assume you have Debit Card or a Credit Card and you have used the same. If you use your card 3-4 times in a span of 10 to 15 minutes, it is more than likely you will get a call from the Bank. The customer support executive will ask you if you have just used your card. The reason this happens is because the system has flagged the transaction as a possible fraudulent case. The basis for flagging the transaction could be basis the fact that historically you don’t have such a spending pattern. On similar lines, one can also leverage these algorithms to identify Fraudulent Claims. A very apt example which I can quote here is identifying cases which could be denied for Medical Necessity. To detail the same, let us say a Patient comes to the hospital complaining of chest pain. The physician checks the patient vitals, asks the nurse to do an ECG and other relevant tests if required. All the data looks normal, but still, the Physician asks the patient to stay in the hospital. The patient gets discharged after four days. In this case, one can understand the patient is kept under observation for one day, but more than that without any supporting information, there are very high chances of the claim getting denied. Other examples where a similar algorithm can be used is the Classification of diseases based on Medical data. One can identify patients with similar issues/symptoms and diagnosis.

Classification is used when we have labeled data, if not, then we use Clustering. So, in case we don’t have Labelled information related to Fraudulent old claims, we can use Clustering to identify similar claims.

c. Natural Language Processing

In the lifecycle of a patient, the hospital captures a lot of information. This information comes in the form of Physician notes, Nurse Notes, Lab & Radiology Reports, and Discharge summary. All of this information together forms Clinical Documentation. This documentation, when in electronic format, can be used in a very powerful way, and we will talk about a couple of use cases. Before a new drug comes in the market, there are a lot of clinical trials which are done. To do clinical trials, one needs to identify and enroll patients for such trials. The process is manual and time-consuming. Identification of Trial Patients is where one can use Clinical Text Mining to mine the Clinical. One such tool is CLIX ENRICH from Clinithink. The second aspect where we can use Text Mining is to identify disease progression. Mining Clinical notes of similar patients, we can gather a lot of information. Clinical Notes mining is possible by leveraging the power of SNOMED-CT. If you are not aware, SNOMED is a collection of medical terms and it has 100,000 plus concepts. Using this and the logic of engrams, we can identify key terms in the Clinical notes. Example, if the note says “Patient comes to the Hospital complaining of Chest Pain,” using SNOMED API’s we can find out if the word “Chest Pain” is a finding, diagnosis or any other aspect. This logic can also be extended to Medical Coding and Hierarchical Condition Coding where Medical Coders have to identify Clinical Terms for billing purposes.

d. Time Series Forecasting

In any business, it is very important to identify the volume of customers expected each day. The same is true in case of a Hospital as well; here is where one can use Time Series Forecasting, to forecast the Hourly Patient Volume. This volume can be forecasted based on prior patient volumes for the past three to four years. In addition to this, one can also leverage seasonal factors like climate to identify the impact on patient volume. Based on the seasonal data, one can also go to the extent of identifying Patient Volume for a specific type of disease. For examples, Emergency Cases/Fractures cases increase during festivals like Janmashtami. On similar lines, burn cases increase during Diwali. All of this information, when tied together, can be used to forecast Patient Volume and accordingly, we can also calculate the Resourcing needs.

e. Survival Analysis

Survival Analysis is generally used for a time to event analysis. This technique is widely used in the manufacturing industry to calculate the warranty period of the product. In healthcare, we can use Survival analysis to calculate the Probability of Survival after the diagnosis of a Disease. As per the Analysis published by IHME, Global Burden of Disease, Cardiovascular Diseases, and Cancer are the Top two causes of deaths in 2017. Cancer has been the top four causes of death since 1990, and as per National Cancer Registry Programme (NCRP), more than 1300 Indians die every day due to cancer. If Cancer is identified at an early stage, the chances of survival increase. In this case, the data which we would require is quite varied; we have listed some of the data elements which can be considered. Age, Gender, Genetic Information (Defects), Skin Type, Location (Geography), Lifestyle (Alcohol, Tobacco, Food), Type of Job/Work. Using this information (Survival Analysis), the physician can decide the medication and diet plan for the patient. Small changes in the lifestyle of a patient can help increase the survival rate of the patient.

f. Deep Learning

Deep learning is a branch of machine learning which can be used to mimic the human brain. There are various Machine Learning Algorithms around Deep Learning, a couple of them being Artificial Neural Network (ANN) and Convolutional Neural Network (CNN).

Diagnostic systems – Neural Networks can be used to detect heart problems; this can be done using ECG information as well as other details like Stress Test. Some of the prominent use cases here are to identify Metastatic Breast Cancer and Skin Cancer. A very common example where Convolutional Neural networks (CNN) are used is in Pneumonia detection. Stanford University has built a 121 Layer CNN to identify 14 different diagnosis just using X-Rays.

Biochemical analysis – Analyze urine and blood samples, as well as tracking glucose levels in people with diabetes, determining ion levels in fluids, and detecting various pathological conditions. 

Drug development – Another key area is the development of drugs for various conditions – working by using large amounts of data to come to conclusions about treatment options.

How can India Leverage Analytics? 

One of the biggest challenges for India is the lack of data(electronic), more than that the means to capture data electronically. Many hospitals do not use any form of a HIS (Hospital Information System) or EHR (Electronic Health Record). If you want to do Analytics, you need data and that too a large amount. Even if we implement a HIS/EHR from tomorrow, we would have to wait for two years to do some decent Analytics. HIS/EHR is a long-term solution, but not something organizations can use immediately. For us to start leveraging the power of Analytics, we need to take the help of Transfer Learning. In simple terms, Transfer Learning is using the research done while solving one problem and applying the same while solving another but related problem. My team.  


Each year lot of deaths happen due to Pneumonia, and the best diagnosis for the same is using a Chest X-Ray. A Chest X-Ray is the most common diagnosis used in any scenario. If we want to start using Analytics in India, one of the areas to focus on can be in the area of Diagnostic Analytics. We can leverage Transfer learning in this area as there are many pre-trained models leveraged by others and available. One example is leveraging ImageNet to identify Pneumonia in Chest X-rays. If one would want to build this from scratch one can even leverage datasets provided by National Institutes of Health (NIH) which has 100,000 Chest X-Rays. In recent times Stanford has released 224,000 Chest X-Rays which can be leveraged for building an improved model.

If you would like to understand more about Healthcare you can download my book “Healthcare Decoded – Begin Your Health IT Journey”.  


1. Retrieved from
2. Comparing Hospital ER Wait Times. (2011, May 2). Retrieved from
3. Burden of Disease. (n.d.). Retrieved from
4. Life Sciences 06/16 Copy. (n.d.). Retrieved from
5. Neural Networks in Healthcare. (2017, April 6). Retrieved from

Harish Rijhwani 
Harish Rijhwani has 17+ years of experience in Healthcare Information Technology. He has worked in Atos Syntel (15 years), Hinduja Global Solutions (1.5 years) and currently in CitiusTech. Provided Solutions across various Healthcare areas Clinical, Revenue Cycle Management, Telemedicine, Analytics, Non-Clinical: HR Payroll, Finance & Supply Chain. Multi-faceted with experience across Solutioning, Pre-sales and Delivery. He has been a Visiting Faculty/Speaker/IT Judge at various events/institutes viz. Welingkar, Somaiya, VIT, Symbiosis, and NASSCOM. He has a passion for teaching Healthcare IT and has done BE Electronics & MBA Systems. He is also author of the book Healthcare Decoded – Begin your Health IT Journey
Read more »

In Discussion with Dr. Thanga Prabhu, @thangas, talks about #HealthIT and the need to share the India Story - The Healthcare IT Experts #Podcast: S1 E5

Listen to the insights from Dr. Thanga Prabhu, on why it's important to tell the India Story for the adoption of #DigitalHealth in India.


The views, thoughts, and opinions expressed in the text belong solely to the author, and not necessarily to the author's employer, organization, committee or other group or individual.

STAY Tuned to the Latest Episodes of the HCITExperts Podcast

Team HCITExperts
Your partner in Digital Health Transformation using innovative and insightful ideas
Read more »

In discussion with Dr. Guriqbal Singh Jaiya on the NDHBIndia Blueprint and Stakeholder discussions - The Healthcare IT Experts #Podcast: S1 E4

Dr. Jaiya, talks about the various approaches that needs to be considered while considering the roadmap of the NDHBIndia Blueprint document and the stakeholder consultations


The views, thoughts, and opinions expressed in the text belong solely to the author, and not necessarily to the author's employer, organization, committee or other group or individual.

STAY Tuned to the Latest Episodes of the HCITExperts Podcast:

Team HCITExperts
Your partner in Digital Health Transformation using innovative and insightful ideas
Read more »

Immersive #HealthTech Ecosystem Showcase in "The Pavilion of the Future" during CAHOTECH 2019 by Manick Rajendran, @manicknj

CAHOTech 2019

Starting from our Pavilion of the Future Evangelists:
TimePort 1 - Dr. Pramod Jacob
TimePort 2 - Abhishek Chaudhary TimePort 3 - Bharat Gera 
TimePort 4 - Dr. Thanga Prabhu / Sundar Gaur
TimePort 5 - Dr. Ravi Modali

These stalwarts took time away from their work to put their mind to creating original creative content and be available on site to present them. 

Aniruddha Nene
Dr. Suman Bhushan Bhattacharya 
Satyam Kumar
Manish Sharma
Dr. Rakesh Biswas

The mentors set aside several tens of hours about at least 100 or more in putting their creative minds to work. 

Not only did the work involve a creative attitude, it involved the dedication of sharp minded attention to detail mindset. It was back-breaking work rising up to a crescendo of sleepless nights towards the finish line.  

1 Fololife
3 Helyxon
5 Shrishti Software
6 Sollu
7 Zbliss Technologies
8 Audicor Cardiometrics
9 dWise
10 Doxper
11 Health Med
12 Raster Images Private Limited
13 ayeAI
14 Infolifetech
15 HealthSensei
16 iMMi Life
18 Kriyatech
19 KraniumHealth
20 Concert Care
21 Vitor Health
22 ViDoc

The Partners and their employees played a pivotal role by being available on Event Day on site and I’m sure several hours of Prep work in the background. I estimate that there must have been an average of 3 people per company who must have put in time. 

The role the Docents played was highly commendable. 

Behind the scenes folks like Ramesh, Karthik who helped create the stalls, Madhava a student doctor who came all the way from Vijayawada and of course our Project Manager, Thilak who fought against all odds to deliver all of the deliverables. 

In all, at least 125 of us put our heart and soul into this event and made it happen. 

Will publish statistics later in the day. 

Thank you all!!

Pre-Conference Information
Hello all, we meet again!

This time we could all meet up at CAHOTech 2019

So, what is CAHOTech 2019?

CAHO is Consortium of Accredited Healthcare Organisations. This is the fourth year they are conducting  CAHOTech, an event of a one-day format with speakers in a conference and stalls by participants. 

Venue: IIT Madras Research Park

Date: September 28

This year, there is a small twist to how we are featuring the companies at this Conference. 

A Brief history of HealthTech Connect-a-thons in India

There is a group of us who have had good success in showcasing the capabilities of HealthTech companies here in India. The first in the series was a Testathon conducted at Philips, Bangalore in 2012. The next was a Connect-A-thon at St. John’s, Bangalore in 2016. The third in line was an Interoperability Showcase at the Medicall Conference in Chennai in the year 2017. 

This year CAHO - 2019 carries the torch.

So What's the Difference in this years Showcase: The "Pavilion of the Future"

The stakeholders in this conference are investors, hospital administrators (CEOs, CTOs, CxOs), physicians, other medical experts and students. To make it meaningful for them, we will be showcasing about 40 companies through “scenarios”. Each scenario while exhibiting specific industry functionality will also highlight four common “competitive parity” themes 

- Customer Experience, 
- Care Delivery, 
- Operational Excellence and 
- Quality

Additionally, we will also showcase industry solutions for two or three Public Health problems in India. This will be futuristic solutions. 

How you can participate and help create this “Immersive HealthTech Ecosystem Showcase”:

1. Register to be part of the Showcase through your company (we will assist you in getting an NSIC grant to cover being part of the showcase)

2. Provide us with a Public Health challenge AND a solution to go with it. 

3. Be there on September 28 to experience what you have helped create.

In addition to the above, those who participate can try their hands at the PitchFest the previous day, 27th September 2019 (they will have to register of course) that brings in a cash purse of up to Rs. 20 lakhs in total. 

Source: Organising Committee - CAHOTECH 2019

We also have several investors committed to being there to check out the offerings.

Here's a look at the conference floor map

What does the The CAHOTech 2019 Immersive Experience have for me as a Startup, as an Investor?

HealthTech companies use technology (databases, applications, mobiles, wearables) to improve the delivery, payment, and/or consumption of care, with the ability to increase the development and commercialization of medicinal products.

Stakeholders consist of hospitals and practitioners; insurance; consumer-facing services; pharmaceuticals; and government. 

The Healthcare industry was valued at US$7.2 trillion in 2015 in the US alone.

In India where the market is huge in numbers and requires innovation to deliver at appropriate price points, the startup community is embracing the challenge.  

For the Startups: 
To showcase what is happening in the Ecosystem, CAHOTech 2019 is providing a platform for startups to exhibit what they do and how they do it in interoperable ways with other companies for the stakeholders. The exhibits will allow the delegate visitor to experience the flow of data and triggers that initiate functionality offered by individual companies in an immersive way through role plays called scenarios. 

These scenarios are designed to illustrate real life situations happening at point of care and the mechanics involved to ensure the delivery of care to the users of the various systems. 

Over and above the core solutions that companies provide to their clients, they also design their systems to ensure a high level of Customer Experience that are delivered in the right manner at the right time for the client at optimal performances. These hidden strengths of the systems designed by the startups will be brought to the fore for the visitor to experience for themselves. 

You will have access to a monitor in your station (we will call it station instead of stall). The scenarios will ensure that you will be able to explain your solution's functionality through a story. What that will mean is, you will present on-screen to the delegates and do your show-and-tell with the gadgets, apps, solutions, demos as well.

The opportunity that CAHO is presenting to HealthTech companies is prestigious and all of you entrepreneurs lend credibility to the event. If you know of partnering companies who are part of your workflow in the ecosystem, please have them contact us. We’d like to showcase all of you. For those of you who have gone through the exercise of constructing a Business Model Canvas (BMC), every one of the entries you made in the Key Partners box is a candidate to be at this immersive experience event. Please refer them.

For the Investors in HealthTech: 
If you are an Investor who wants to see the potential of the healthcare industry, or if you are a Healthcare Organisation Administrator who wants to explore how you can put together various tech products to secure your place in the marketplace or if you are a Clinician and would like to know how these technology systems work in concert, or if you are a student who wants to know your career prospects or if you are any kind of stakeholder in healthcare and would like to get a touch and feel of what the industry is about, you need to be here.

Message from, Sameer Mehta, Dr Mehta’s Hospitals, the Atlas Family Office, the Chennai Angels (EC) & Organising Chair -CAHOTech, Vice President - CAHO
"Welcome to our CAHO Community.  We hope many of you are excited to be part in the first ever Health immersion program in India.  This will be a wonderful program - the more you put in, the more you will benefit.  Spread the word.  Your prospective customers and investors should experience this at CAHOTech...
With over 500 folks expected and over 10 investors committed - this promises to be interesting."

How can I be a part of the Showcase?

Please fill the CAHOTECH 2019 - Showcase Registration Form

Alternatively you can download the WORD version of the form - here

Any queries about the form, please Send in a email to: [email protected]. You can also leave your comments to the article and the Team HCITExperts will be able to get back to you. 

Some FAQ on the Form:

Some of the questions we have received have been:

- can the primary problem statement be more than 1 (# G)
G. Yes, they can be more than 1

- what do we highlight as our Strengths? (# I)?
I. Every aspect of your product that you feel are strong from the point of view of your customer 

- what do we include in Operational Excellence (# M)?
M. Factors that contribute to the better operation of the facility of your client

- what do we include in Quality of Care (# N)?
N. Factors that contribute to the quality of care experienced by the patients; this is different from Customer Experience which is more to do with the usability of your product. 

Bottom line, we want to showcase your offering through an experiential way. The more information you give us and indication of what you want highlighted, the better for us to work on your scenarios.

Manick Rajendran
Founder at iMMi Life
Read more »


Popular Posts