Title: painting of Isaac Newton
Artist: William Blake
Location: Tate Modern, London, UK
Dimensions: 460 x 600 cm
“The tree which moves some to tears of joy is in the eyes of others only a green thing that stands in the way. Some see nature all ridicule and deformity… and some scarce see nature at all. But to the eyes of the man of imagination, nature is imagination itself. ” William Blake
Of course the quote in the title belongs to one of my all-time favourite authors, the amazing Aldous Huxley from his book “Note on Dogma”. At the end of the day, this quote represents why our product roadmap points directly in the direction of big data medical imaging analytics and we are adding features in analytics.3dnetmedical.com faster than the speed of light.
But why predictive analytics in medical imaging you may ask?
Let’s consider the evolutionary steps of medical imaging after the 90’s.
Phase 1 – The golden era of PACS: At the beginning digital medical imaging, strictly the new toy of the radiology department became an indispensable tool for modern healthcare. We started seeing more images, richer images, performed more often. This was the heroic period of medical imaging, the transition from film to filmless, and the focus was on generating and interpreting digital content.
Phase 2 – The era of enterprise imaging: In phase 2 we realised that the same digital content has more value than what is needed for a radiologist to perform and be efficient. Other healthcare professionals needed to see them, either in the same building or at a remote location. And to make it even more interesting, it wasn’t just radiology anymore. Many other healthcare disciplines started to generate their own stream of rich imaging data contributing to the digital information obesity. And I am not talking here only about the cardiology department next door, I am talking doctors very often located at remote locations acquiring images. It is my prediction that the amount of data generated by these other disciplines will overcome the modest needs of the radiology function within this year. Many companies refer to this phase as the era of enterprise imaging and name their offerings appropriately. Perhaps VNA is the illegitimate heir of that era (more on that I hope in another blog)
Phase 3 – when Big Data changed Radiology: And then increasingly healthcare managers and other clinical professionals (even academics) realised that there is a treasure trove of extremely valuable data embedded in those images and associated information; and they collectively raised their hands and requested easy access and a slice of that extreme digital consumption. And soon everybody got in the game. Regulators and auditors asked for access, Insurance companies and other players wanted access, and of course the well informed patient, amply educated by Dr. Google asked nicely if he could please have access. And one day all of a sudden, Radiology is not just about capturing and interpreting images anymore; it is not just about making sure that images are shared in the Enterprise any more. It is about providing different facets of valuable information well hidden within the Big Data Imaging archives, facets based on how you can use them and benefit from them. Welcome to the brave new world and the era of Big Data Analytics.
Because this is what big data medical imaging is all about. Multiple medical imaging producers and multiple consumers of extremely valuable content, difficult to manage using traditional PACS systems. And every time traditional PACS tried to provide an answer to the questions all those varied stakeholders asked, it increasingly felt like being given a car without anyone telling them how to drive it, and without a road map, increasingly driving blind. A car not fit for purpose. That’s a virtual anarchy in medical imaging.
Analytics.3dnetmedical.com will provide the answer to that. Because it is with predictive analytics that we will see the true and essential point in this anarchy of never ending information overload, and we will leave the rest as surplus. Albert Einstein once said that 90% of the data we collect will never be actionable or even helpful, and then he admitted that he made that up because he could not actually measure it. Moreover, what we really hope to offer with analytics.3dnetmendical.com is Actionable Analytics, or the discovery and communication of meaningful patterns in medical imaging data that hopefully lead to actions. Ideally those actions should be predictive, shining a light to the future, focusing on the big picture.
The value of Medical Imaging Analytics: “Sustaining high clinical and business performance is a product of continuous strategic alignment.”
The writing is on the wall. Analytics is fast emerging as the “next big thing” in health IT. In Medical imaging, the main culprit for the production of more than 95% of health IT data, everybody realises that analytics is even more important. 3Dnet customers are asking at an increasingly and alarming rate for advanced tools to get valuable insights to the hidden information wihtin their medical imaging repositories and use that not only to streamline their work and improve efficiencies, but more importantly to help them with their growth strategies whilst improving their clinical outcomes.
Let me please tell you this story. A CEO of one of our large clients with a couple of hundred imaging clinics under his management told me recently: ”The monthly reports that I get from my guys are of not use to me anymore. I want real time dashboards to provide virtually immediate feedback on my operations. This is absolutely essential to have information at my fingertips and be able to take critical decisions for the growth of my company”. What he was demanding from us, as his technology partners, was to be a big part of his internal digital strategy by providing to him the business analytics he needs and deserves. He wanted to get better and real-time insights of his equipment utilisation and determine each clinic’s efficiency. For example “Is one machine used a lot?” “Are patient wait a long time?” He also demands better insights in personnel utilisation. “How busy his radiologists are truly are?” “Can he come with better ways of utilising their rare and expansive skills?” “Can he merge or blend services to improve the overall outcome?”
We went for dinner and we started talking about all that, the future of medical imaging analytics and the implementation challenges we are facing. We had with us his Chief of Clinical Operations, a very prominent Radiologists and opinion leader. As the discussion was progressing we realised that it is only with analytics that we can provide a solution to one of the most interesting problems of modern medical imaging; that of lack of proper quality control mechanisms and the overall drop in quality of medical imaging. The proliferation of medical imaging outside radiology, combined with the introduction of new and very complicated protocols and scanners and the increasing realisation of lack of the wider availability of proper skills to deal with that amplifies the problem of quality, to a point where it is maybe one of the biggest threats in modern Radiology. Interestingly enough, it is not only us that outline that as a major risk, it is also the American College of Radiology and RSNA in USA and Royal College of Radiology in the UK. Quality control in medical can be addressed either the inefficient way (by manually auditing a random 10% of examinations as NHS recommends but has no resources available to implement), or the efficient way with automated, real time predictive analytics that number-crunch all the examinations of the repository. Our company with 3dNet analytics has embarked in a journey to address that. As Radiology Services are entering the commodity stage, this is the only way to truly de-commoditise this market.
Peter Drucker, one of the gurus of modern management, pioneered the concept of “Management By Objectives,” which shifts the focus from process to goals and to the purpose of the activity rather the activity itself. I believe that this concept will form the new cornerstone of this evolutionary phase of medical image management systems and processes It even has the potential to offer a realistic solution to the problem of medical imaging obesity. Instead of asking, “What do I do?” the medical imaging business and clinical professional will ask, “What is the objective toward which I am working and how do I measure it?”. Analytics.3dnetmedical.com will enable clinicians to use medical imaging do five things really well:
- organize and structure the whole range of available medical imaging data and knowledge,
- test these structures,
- predict behaviour, and test “what if” scenaria
- gauge the soundness of decisions and clinical diagnosis in disease management,
- and analyse and improve the performance of the healthcare provision when it comes to medical imaging services.
Inevitably nobody can stop the clinical knowledge progress in the rapid evolutionary phases of medical imaging. If doctors cannot survive all the information the data producers with increasingly create, we are in serious trouble. No one is going to stop creating information, contributing to the medical imaging obesity problem. The quest for the cure of Clinical Nerve Attenuation Syndrome is more urgent than ever, and only medical imaging analytics can provide a sustainable and viable cure.
I think ten years from now, when we look back at how this era of big data in medical imaging has evolved, we will be stunned at how uninformed we used to be when we made business and clinical decisions.
Analytics at analytics.3dnetmedical.com.
This is very interesting stuff and combined with your other posts a valuable (-external) insight into the future of medical imaging practice and radiology. Part of me can’t help feel that the entity lost in this institution/population level of analysis is a) the patient and b) the carer. Surely we will generate very interesting statistics that can inform healthcare choices and necessary efficiencies required at national level. There is another (by no means perpendicular) model though that basically says “let the experts do their thing” an example is mentioned here:
Which has seen huge gains in efficiency, job satisfaction, levels of care and cost. Maybe the tech should support these efforts rather than layers of data munging for insurance companies and managers to pour over in the quest for spread sheet efficiency? I am very much persuaded by your previous talks on patient empowered imaging in this respect.
Thank you for sharing your day to day insights.
Ron Marrocco said:
Data analytics can be a great way to understand what is happening within a population or within a healthcare system. It will not be useful, however, if the data that is needed to answer the question being asked is not accurate, captured in a timely manner or querried appropriately. Systems that capture healthcare information should ‘begin with the end in mind’ in order to be of value.
This article generated an interesting discussion on Linkedin forums. I copy some of the comments below:
Amalia Couselo said:
“Peter Drucker, one of the gurus of modern management, pioneered the concept of “Management By Objectives,” which shifts the focus from process to goals and to the purpose of the activity rather the activity itself.”
Narayanachar Murali said:
Until analysts learn to look beyond the CCHIT EMRs as the data source for big data, it is unlikely we will find any use for analysis of big data. When you create huge silos of forced data entry tied to billing and compliance with taxpayer money at stake, you get meaningless garbage which does not reflect clinical realities. To this end big data analysis has failed “health care”.
Big data is certainly helpful in fraud detection and medicare is doing a good job at it by analyzing billing data, focusing energy on extreme outliers bilking the system. I wish it could call halt at the time of utilization( just as credit card companies look at heuristics and place transaction hold). If data analysis does not result in rapid policy changes it helps no one
Dominique du Crest said:
Ronald thank you very much for sharing this article. My center or interest is skin from beauty to disease. As far skin imaging there are still major unmet needs. Concerning data analytics specific to skin, do you have infos to share?
Ronald D. Marrocco said:
Big Data can pack a big punch if we ‘begin with the end in mind’ as we determine what we need to know and why. Without a clear definition of what we want to know and what kind of data we will need to answer the question, disappointing and inaccurate information is a real possibility (and sadly the likely outcome). In setting out to capture data, always ask, “What do I want to know and am I able to capture key inputs (accurately) in order to draw useful conclusions and inform decision-making?”
Amalia Couselo said:
This is the reality¡¡
A CEO of one of our large clients with a couple of hundred imaging clinics under his management told me recently: ”The monthly reports that I get from my guys are of not use to me anymore. I want real time dashboards to provide virtually immediate feedback on my operations. This is absolutely essential to have information at my fingertips and be able to take critical decisions for the growth of my company”. What he was demanding from us, as his technology partners, was to be a big part of his internal digital strategy by providing to him the business analytics he needs and deserves. He wanted to get better and real-time insights of his equipment utilisation and determine each clinic’s efficiency. For example “Is one machine used a lot?” “Are patient wait a long time?” He also demands better insights in personnel utilisation. “How busy his radiologists are truly are?” “Can he come with better ways of utilising their rare and expansive skills?” “Can he merge or blend services to improve the overall outcome?”
Alessandro Mazzarisi said:
Really do you think healthcare organizations willl share for free their streams of data concerning managent, workflow and scheduling, letting others to take independent decisions? We all, are very sensitive about patients privacy, so why every chief of healthcare departments should gift to big data analitic processors their faults and inner problems? Apologise for my misknowledge about USA market. I’m part of EU-IT healthcare System. Here helthcare organizations are not armonized yet for electronic patient records among regions and contries yet. Actions to grant interoperability between devices and organizations are strictly under regional regulatory authorities that mess up managers and stakeholders. I think big data analitics techniques are very well in other fields, not for healthcare organizations.
Cary Oberije said:
Big data can really advance the medical field, but (as always) there are challenges. Different hospital systems are being used, making it difficult to create big databases. Moreover, definition of variables is usually not comparable or data is unstructured text. And very important: privacy has to be ensured. In the EuroCAT project we developed a great solution to be able to share data while preserving privacy. For everybody who’s interested, please take a look at our animationhttps://www.youtube.com/watch?
Donald Bryant said:
And then there is “small data analytics”. Using local data at a site to improve the patient outcomes is possible with a focus on population level health management. I have found analyzing local, population level outcomes data coupled with measurable objectives can lead to improved outcomes for both provider and patient.
Ronald Greeff said:
Ronald and Cary – There are 3 aspects to our analytics engine – the first is collection of data. Different systems, different formats, different naming conventions is a big problem in healthcare that is compounded when you’re talking about distributed healthcare networks. Using the DICOM and HL7 standards, we’re able to (in the first instance) tell you what your data looks like and tag it all.
Once it’s all tagged, the question then becomes “well, what information do I want” – and the answer is likely to change as the organisation changes. Therefore it isn’t feasible to “pre-configure” databases. For that reason, we’re using leading non-relational databases.
The questions about privacy are of minor concern. For example, you don’t need “patient name” in order to compare the incidence of ischemic heart disease between two geographic regions, or any kind of patient identifier if you’re looking at “delta time between patient arrive at clinic – patient scanned”. The data you extract is at a higher level. I can query PACS for instance and say “how many patients came in for a CT scan at clinic A”, get a value, then compare that value to the same one at another clinic.
Jay Kamdar said:
Nice one Ronald.
Some great insights towards a realistic 21st century era of Personalised Medicine is what this leads to. The value of Big Data analytics is immense typically for cash strapped NHS trusts like mine who could leverage on the operational efficiency arm from processed data and save a big chunk on meaningless outdated processes with zero yield.
On the private front, innovative healthcare providers like HCA in Central London and U.S. can use these analytics to develop best practices tailored for best outcomes especially in cancer care and cardiovascular / chronic disease managements. Value would be immense in the long run, agree with Cary & Alessandro on the confidentiality aspect though. But I see this as a great asset in the London private providers toolkit of diversification and a marriage of IT to healthcare in the coming years if it delivers on the promise. Can it become the core to NHS nsustainability ? I think it will ….