© istockphoto.com/metamorworks

“One decisive factor for data analytics projects to succeed is combining artificial and human intelligence”

An interview with Dr. Philipp Liedl, Managing Director of STASA Steinbeis Angewandte Systemanalyse GmbH

We are surrounded by data, at work and at home. Filtering out information that is actually relevant and using it properly is one of the biggest challenges of modern times. Naturally, the quality of data is crucial. One person who knows a lot about this issue is Steinbeis expert Dr. Philipp Liedl, who talked to TRANSFER magazine recently about using AI for data analysis, big data, and applied systems analysis in the field of medicine.

Hello Dr. Liedl. Your Steinbeis Enterprise converts data into knowledge. What do you see as the biggest challenges at the moment given the increasing flood of data and information?

One of the greatest challenges is to extract the information that is actually relevant from all the data that’s available. One way to do this is to use AI methods to analyze data. But it’s important to pick the right input variables – or features – when you do this to go into the AI models. If you don’t, analyzing the data ends up raising questions rather than answering them. To do this, you need to draw on expert knowledge in each field of application. This makes the process of spotting mutual relationships between different influences more reliable and the results can be interpreted more quickly. We’ve found that one decisive factor for data analytics projects to succeed is combining AI algorithms with expert knowledge provided by humans, so you have to combine artificial and human intelligence.

The increasing volumes of data now also make it more difficult to check data quality. Not only does this apply to information on technological processes, things are even harder with data in the field of social sciences. This increasingly means you have to use automated testing methods to safeguard the required data quality – for example, to identify data blips or data errors.

How important is it to think laterally across different disciplines in managing the flood of information?

It does help to use methods across different methods to process and analyze the flood of data, and it can also help if you filter out the information that is actually relevant with an eye to reducing data complexity. For example, at STASA we’ve succeeded in adopting algorithms – which we originally developed for analyzing mass data from manufacturing processes, including the variables they produce – and transferring these to other application areas so they can be used to extract mass data characteristics.

We’ve used our interdisciplinary approach to develop a number of best practice methods in recent years, and these have enabled us to deliver a variety of successful projects relating to data analysis, modeling, and forecasting – for customers in industry, trade, and the public sector. Our aim is to work closely with customers so we always understand the specific nature of each particular application within our solutions. We also want to forge links between different disciplines by talking to our customers within a broad variety of specialist fields. This helps ensure that implementing data analytics projects with the customer results in a win-win-situation.

What potential do you see in using big data methods in the field of medicine – but also what risks does it present? And what impact do these have on your work?

Our work in the healthcare market involves analyzing regional data and developing software tools. From that perspective, there’s significant potential in strengthening networks between different healthcare services. This can be done by providing a platform for health-related topics, such as telehealth solutions and home care.

In the future, demographic change will result in increasing demand for healthcare services because society is going to get older and older. The demands will be different depending on the region, and there’ll be a particularly strong rise in older people in rural areas, who’ll be overrepresented because lots of younger people will wander off to the urban areas. It’s also more difficult to provide healthcare services in rural areas because you have to travel longer distances to get to a doctor. Also, in sparsely populated areas, providing nursing services in the home involves long driving distances, so overall, care workers have a greater time investment than in urban areas. Telemedicine services or platforms for improving the organization of healthcare services and nursing can help in this regard by meeting the growing demand.

At the same time, working with digital media becomes difficult for older people and they’re less proficient in it. As a result, the solutions provided in this area have to be even more user-friendly than in other areas and offer intuitive controls. In terms of risk, one of the main factors in this area is data protection. You have to gain the trust of users. So it’s extremely important to adhere to prevailing data protection guidelines. Also, such platforms should only ask users for information that is really needed to provide the online services. The responsibility lies with us as developers in this respect, although the operators of such platforms also bear a responsibility.

You often use methods of applied systems analysis for your work. Can these also be applied to public health services?

It’s easy to transfer the methods of systems analysis to the healthcare system and it’s already being used there now. We also apply our methods to the public healthcare system, for example by merging our models of local-area demographic development with the issues faced in healthcare.

This is where the increasing use of digital solutions in the healthcare system helps. For example, we’re currently working on a project called DiCaSA, which stands for Digital Care Supply Advisor. It’s backed by NBank with funding through the European Social Fund and involves developing a web-based platform, which among other things should improve care provision at home in rural areas. The platform will allow people requiring care and their families to link up with care providers in such a way that it will not only be possible for them to find the best possible kind of care options, but also care workers will be able to improve their services by planning the coordination of appointments. This should improve care worker capacity and ensure it’s put to optimal use for people requiring care.

There are also a number of other interesting ideas. For the providers of healthcare services, such as hospitals, physicians, and pharmacies, having graphical displays of patient characteristics in geographical terms, so on maps, and anonymously of course, can be really useful for supply and demand planning. For example, they can be shown interactive maps that indicate postal areas at a glance where there are particularly high or low numbers of certain patients.

Enriching your own data with other data – of a socio-economic nature, such as population numbers by age group or gender – allows you to gain quick insights, and these can be used for demand planning. For patients, it would be interesting to provide platforms that map the availability of services in the healthcare system, so for example they could find out where the nearest doctor or pharmacy is and work out the best way to get there on foot, by car or by public transport.

Lots of us are currently following the number of COVID-19 cases on interactive maps in the internet provided by different research institutes and the media. In geographical epidemiology, it’s been common practice to conduct such analyses on the transmission patterns of epidemics for years now.


Dr. Philipp Liedl (author)
Managing Director
STASA Steinbeis Angewandte Systemanalyse GmbH (Stuttgart)