Data Analytics For Healthcare

Data Analytics for Healthcare

In healthcare, the overarching aim is to improve patient safety, recovery and outcomes. All processes, activities and technologies that are carried out to support this aim generate data. So every healthcare business knowingly and unknowingly amasses a wealth of data; but the business is able to gain an edge over competitors only when this data is broken down and analyzed to uncover hidden insights.

Big data and analytics are no longer new, flashy concepts. For every company wondering how to use its data better, there are (literally) hundreds of technology suppliers claiming proven capabilities in extracting and analyzing data. So why does much of data analytics still remain a big mystery? Why isn’t every healthcare business able to start using its own data more effectively (without having to invest in outside vendors)?

Data analytics 101

Your checklist for analyzing data in-house

4 examples of data analytics in healthcare

Cheat sheet for building a data analytics solution in-house

Data analytics 101

  1. Data extraction and management

    Almost every process and technology that your business deploys creates data. This data is usually stored in semi-structured and unstructured sources (such as images, audios, videos, email text etc.). These are unstructured because they are difficult for a computerized machine to interpret or discover patterns in. This data needs to be mined, validated and standardized (in other words, structured) so it can be further used in meaningful ways.

  2. Descriptive and diagnostic analytics

    These practices in analytics use your current data to provide insights and past data to provide hindsight. To achieve this, data analysis methods such as exploratory analysis, correlation, clustering and trend analysis are used. The result? Segmentable dashboards and reports for the senior management and operational teams to make better, more effective decisions. An excellent example of diagnostic analytics is bringing up/flagging exception cases hiding in the current order volume. This helps the company improve the way it approaches order fulfilment and improve performance. Technologies like Microsoft Power BI, SAP Lumira, Tableau and Spotfire are commonly used for descriptive and diagnostic analytics.

  3. Predictive and prescriptive analytics

    If you want to step up your game and get a much deeper understanding of customer behavior and intent, predictive and prescriptive analytics are the way to go. These practices are often supported by artificial intelligence/machine learning and include advanced data analysis methods such as regression, neural networks and support vectors. The result? They predict future trends (that are relevant to your KPIs), prescribe the best course of action and enable you to take preventive actions. Technologies like Azure Machine Learning, Python, Apache Spark and SAP Hana are often used for predictive and prescriptive analytics.

Your checklist for analyzing data in-house

If you want to start analyzing your data in-house, you’ll need to make sure you tick these 3 boxes:

  1. Understand your business

    Obviously, you know the ins and outs of your business, but you also need to know the KPIs (key performance indicators) that are most important to your senior management at this point in time. If you’re a pharmaceutical manufacturer/distributor, are order completion and inventory optimization your most valued KPIs? If you’re a hospital, is patient wait time reduction a goal you’re prioritizing in the short term? If you’re a clinical lab providing services to clients across geographies, are you looking to keep tabs on your current projects across multiple locations? You get the idea.

  2. Understand your data requirements

    You also need to fully understand what it is that you want to accomplish with your data (in other words, ‘your data requirements’). These may or may not be different from your overall business objectives and KPIs. For example; If your business goal is to increase the volume of pharmaceutical/medical device orders you successfully complete each month, your data requirement may be to bring up exception cases hiding in your systems. This way, you’ll be able to view unfulfilled/abandoned orders at a glance and take actions to complete them.

  3. Build a decision support system (DSS) to satisfy those requirements

    Basically, this DSS is your data analytics software that provides business intelligence reports to your internal teams and senior management. It extracts and analyzes your data and picks only the relevant, simplified insights to display through drill-down dashboards/reports.

Great theory, but what about practical applications of data analytics in healthcare?

Here’s how 4 real-life healthcare businesses are successfully analyzing and using their own data:

  1. A medico-legal service provider in the U.S. :

Situation: As the nation’s leading provider of medical record summarization, record retrieval and litigation support services, this company serves over 3,000 law firms and 500 insurance carriers worldwide.

Challenge: The record retrieval and summarization business is highly process-oriented and depends on records being released and vetted by authorized medical professionals within rigid timeframes. With such a massive volume of medical records to request, summarize, manage and customize (for individualized requests from customers), this company struggled to satisfy every customer order. Orders were getting stuck and lost at key stages during the process. What’s more, the senior management was unable to identify where these problems occurred.

Solution: Order and project analytics solution:

  • Real-time, intuitive dashboards and reports for senior managers that provide 360° visibility on order status, escalated orders and total requested vs completed orders. These reports can be further filtered by geography, age, team and service line. This enables the company to quickly notice ‘red flag’ orders; the company can then prioritize these orders to ensure their timely completion.
  • Analysis reports on order age, key milestones within the order lifecycle and the average time taken to reach each milestone. This helps the company visualize the stages within their process that the delays occur in; it’s super easy to then find ways to optimize the existing process!

  1. A Fortune 500 pharmacy network in the U.S.:

Situation: This pharmacy network supplies numerous pharmaceutical drugs to hospitals, retail pharmacies and specialist physician practices every week. The network uses a range of courier vendors for this distribution cycle.

Challenge: The pharmacy network struggled with ensuring that hundreds of custom pharmaceutical orders reached the right customers, at the right time. As many of their orders are stat orders, the network needed to gain visibility on the current delivery processes and find areas for improvement. Also, there was no way to verify whether the payments made to courier vendors accurately matched the services provided by them.

Solution: Supply chain and freight management analytics:

  • An order management application that extracts and analyzes all current and upcoming stat orders, courier vendors (including the cost of hiring that vendor) and the delivery cycle for each order. This provides the network complete visibility on the effectiveness of current supply chain operations and led to savings of $15.5 million/year in parcel and courier spend.
  • A freight management feature that consolidates courier invoices, contracts with courier vendors and Advanced Shipping Notice (ASN) records. This application then measures the invoices against the contracts; factoring in any service dissatisfactions caused by courier vendors. The resulting invoice audits and accounts payable reports led to 25-30% cost savings in courier charges.

  1. National Health Service (NHS) in the UK:

Situation: NHS is the U.K.’s biggest publicly-funded cluster of healthcare organizations. NHS England in particular, commissions primary care services (GPs), specialist services and allocates funding to 211 regional clinical commissioning groups across England. NHS also judges, manages and awards funding to clinical research and development (R&D) projects across the U.K.

Challenge: As medical/clinical R&D is a highly competitive and dynamic area, NHS aimed to grant funding to only the most relevant, feasible and implementable research proposals. To this end, NHS struggled to evaluate the actual expenses of current R&D projects against projected benefits. Also, the senior NHS management had no visibility of prioritized R&D projects.

Solution: Innovation scorecard application:

  • A two-way interface that enables innovators of medical devices/services to apply for funding and enables NHS authorities to measure project expenditures against allocated budgets.
  • A reporting dashboard that enables NHS directors to understand and control long-term projects supporting medical device development across the UK.
  • The scorecard assessed 300+ R&D proposals and saved approximately $15 million over 4 years!

  1. A hospital network in the U.S.:

Situation and Challenge: This network managed several hospitals across the county. Due to the recent shift from ‘fee-for-service’ to ‘value-based payments’, this hospital wanted to accurately understand and measure the services provided by its physicians.

Solution: Physician practice analytics application:

  • Extracts and analyzes data on the number of physician visits and physician charges; based on this, a Health Information Network (HIN) was built for the entire county.
  • A centralized view of physician operations across the county helps the network gauge and improve physicians’ productivity.

Cheat sheet for building a data analytics solution in-house

So you love the many use cases of data analytics and want to explore how this could help your business. Great! But how exactly do you go about deploying a data analytics software? Build it in-house or outsource it? At this point, many tech vendors support the case for outsourcing. Not this one though!

Obviously, we’d love the chance to partner with you and work as your decision support system, but we understand that you first need to objectively assess if you can build an analytics solution in-house. Download our comprehensive cheat sheet below to learn how you can build this software yourself:

  1. Conduct a requirement analysis

    Think about your current data journey, how do you currently use your data? Do you want to track orders, manage your supply chain or assess the efficiency of your HR assets? Do your operations/project teams manually track data? Do you already have a BI software that falls short of the expectations? Does it just focus on analyzing your past operations whereas you want future predictions and suggestions? Are there outdated interfaces, poor usability or complex integrations with existing systems? Also consider if you want to get more out of an already-existing data-heavy system (such as your ERP/CRM) by adding a data analysis software, or if you want to build a fully custom core application.

To find answers to these questions, interview internal target users (who are responsible for managing/using these processes and projects every day) and identify their pain points. This will help you to create a list of “must-have”, “good-to-have” and “not necessary” features and get a direction for designing your BI software.


  1. Determine project scope – Creating a wish-list of desirable features and functionalities is great, but which of these features are the most relevant, most important and most feasible for you to implement? Below are some features you may want to include in your project scope:

  • Integrations with existing systems –Before any analysis work starts, data needs to extracted, cleansed and organized from multiple data sources, types, applications and APIs. The ability of your data analytics software to seamlessly blend with your existing systems (SAP, ERP, CRM, accounting databases, legacy systems etc.) is crucial here. The technology community runs rampant with stories of analytics software that do not integrate with the company’s internal applications; thus being unable to extract the data that the software is designed to analyze in the first place! In such cases, the company needs to manually extract and feed data into the software, which is an example of cumbersome inefficiency that you want to avoid!
  • Data exploration and scalability – In order to build the analytical algorithms and models that form the foundation of your analytics software, you would undergo a data discovery and exploration phase. This would help you understand the business impact of the problem you are trying to solve and frame analytical questions accordingly.But besides doing data exploration merely to build this software, you may want to include some exploration functionalities in the software itself. These may include streamlining the testing of new hypotheses and helping to eliminate the bad ones faster. This may also help you to build scalability; i.e. the ability to test different data models on large data sets quickly.
  • Drill-down and roll-up – The reporting dashboard that your analytics software generates should allow users to drill down to greater levels of details or roll up for a broader, bird’s-eye view. For example; if the percentage of completed orders is down over the last quarter, you want to be able to dive deeper and see where the delivery is coming up short. You’ll also want the ability to apply filters to view particular segments of the entire picture, such as total completed orders for the Northeast or the interaction history with a specific customer.
  • Customizations – Based on your requirement analysis, your analytics software will offer predefined reports and dashboards that solve the needs of the particular teams you are trying to help. But healthcare business is ever-so-changing, and you want to empower your internal teams to do real-time in-depth analysis on the fly to discover insights and make decisions. These teams should not have to come to you every time they want to tweak their analysis or try out a new variable. The software should walk them through the steps to running advanced ad-hoc analyses like forecasts, trend analysis, pattern recognition, clusters, gap analysis, comparisons etc.
  • Responsive design – Besides the above features, a responsive design (that works both for desktop and mobile platforms) should definitely be part of your project scope.


  1. Decide the technology architecture, maintenance and support

    Once you’re crystal clear on what your analytics software will include, you need to decide the technical and functional framework for it. How will the data be managed and standardized? (Hint: creating a structured data warehouse will ensure data continuity and standardization). What technologies will be used to build each module and functionality of the software? What hardware will be required to support this? How will possible issues, changes and configurations to the software be managed? Is there an internal team available to maintain the software, support and train the users? Some key factors to remember when considering technologies and data servers are: server performance, availability, scalability, maintainability, recoverability and existing infrastructure.

  1. Plan the project

    Now, you need to actually plan the project. We recommend following the principles recommended by the Project Management Body of Knowledge (PMBOK)! Below is our tried-and-tested project management framework that we’ve used for 20+ years!

You should start by mapping out the detailed project phases, key milestones, timelines, relevant stakeholders and the tools needed to complete each phase. The following tools/concepts are amazing to develop a comprehensive, unambiguous project plan:

  • Responsibility matrix (RACI)
  • Communication matrix
  • Project tracking and monitoring matrix
  • Risk management and change management contingencies
  • Quality assurance process

  1. Initiate the project

    Once you get the senior management’s alignment, all there’s left to do is execute the project, keep everyone in the loop and track progress. Make sure to follow the Agile and Scrum frameworks with 2 week sprints; this way you’ll receive constant iterative feedback from the target user which you can immediately apply to enrich your product development. At the end of it, you’ll have your own fantastic data analytics solution!

If you’d rather focus on your core activities and leave the technical stuff to us, we’d love the chance to work with you as your decision support system and build a data analytics software completely suited to your exact requirements! Our experience analyzing data for 7 healthcare clients over the years empowers us to adapt to your needs and serve you!