Steffen Baumann, PhD

AI in Radiology: To Trust or Not to Trust AI? A New Frontier in Diagnostic Accuracy 👇 Alexis Ducarouge, GLEAMER’s CTO, recently shared an intriguing case where their chest radiograph analysis software identified a lesion hidden behind the heart—something not clearly visible to many radiologists on a unique standard chest X-ray incidence. The lesion, however, was confirmed on a chest CT scan. This case underscores a pivotal moment in AI development: AI systems are now capable of detecting abnormalities that might elude human radiologists. Yet, the tendency to dismiss AI findings as false positives if they conflict with human judgment is concerning. We are at a critical juncture where machines outperform humans in specific tasks. AI trained on robust datasets, including both 2D radiographs and 3D chest CTs, can detect subtle anomalies that may go unnoticed by human eyes. This presents radiologists with a challenge: to trust or not to trust AI? As a radiologist who embraces technological advancements to enhance diagnostic accuracy and patient care, I carefully consider all AI-generated data. Experience has taught me that AI can be correct even when my initial assessment suggests otherwise. Caution is necessary in both directions, and the implications of AI findings must be thoughtfully considered. Integrating AI into our diagnostic process doesn’t mean blindly trusting or dismissing it but incorporating it as an additional data stream to enhance our clinical judgment. In this instance, verifying AI results with a CT scan was straightforward. Future challenges will arise when AI provides insights that current diagnostic tools cannot verify. How we adapt to and integrate these advancements will shape the future of radiology and medicine. For more details, check out Alexis Duracouge's original post in the comments below, and feel free to comment and challenge this point of vue ! #AI #Radiology #Trust GLEAMER #ChestView

82

8 Comments

Integrating digital health devices and technologies in clinical trials is revolutionizing data collection, analysis, and utilization, enhancing trial efficiency and patient engagement. Recently, I explored this topic with experts from Masimo and Medable, Inc. Key Insights: - Enhanced Data Collection: Integration of Masimo’s devices with Medable’s platform via Bluetooth connectivity improves the ease and accuracy of data collection and transmission. - Improved Data Quality: Masimo's advanced sensor technology ensures high-accuracy data capture, even under challenging conditions like motion artifacts and diverse skin tones. - Real-Time Monitoring: Continuous vital signs monitoring, especially in oncology trials, enables early detection of patient deterioration or adverse reactions. - Digital Biomarkers: Utilizing digital biomarkers such as physical activity levels and heart rate variability provides real-time insights into disease progression and treatment response. - Regulatory Challenges: Navigating the regulatory landscape for medical device compliance is complex yet crucial for successful real-world data integration. - Future Advancements: AI and advanced algorithms will significantly enhance data analysis and treatment personalization. Integrating these technologies streamlines clinical trials, promoting a patient-centric approach, improving compliance and data quality. The future of clinical trials looks promising with these innovations. #DigitalHealth #ClinicalTrials #MedTech #HealthTech #PatientEngagement #DataAnalytics #AIinHealthcare #WearableTech #RemoteMonitoring #HealthcareInnovation #Medable #Masimo Musaddiq Khan Daniel Cantillon M.D. Lisa Barbadora

12

👉 The Accu-Chek® SmartGuide CGM solution provides 14 days of accurate real-time glucose values for adults living with type 1 and type 2 diabetes on flexible insulin therapy. 👉 The built-in AI-trained algorithms will empower users to proactively intervene when their glucose levels require attention and before a complication can even occur. 👉 The solution is set to launch in selected European markets in the coming months. Roche announced today that it has received the CE Mark for its Accu-Chek SmartGuide® continuous glucose monitoring (CGM) solution. This significant milestone paves the way for the solution to be made available to people living with type 1 and type 2 diabetes over the age of 18 on flexible insulin therapy. “Maintaining optimal blood glucose levels and preventing adverse glycaemic episodes remains a complex task for people living with diabetes, often necessitating up to 180 therapy decisions a day,” said Matt Sause, CEO of Roche Diagnostics. “Our novel CGM solution with its predictive algorithms will help address significant unmet needs associated with diabetes management, empowering users to take control of their condition and live better and healthier lives.” Despite CGM technology's demonstrated positive impact on glycaemic control, a significant proportion of people living with diabetes still do not meet glycaemic targets - even when using a CGM system. Moreover, they typically encounter an average of two hypoglycaemic episodes a week, with 1-2 of these being severe enough to require medical intervention each year. Notably, nighttime hypoglycaemia is associated with reduced quality of life, increased anxiety, and fear. The persistent fear of hypoglycaemia, hypoglycaemia unawareness, sleep disruption, and diabetes-related distress among CGM users frequently correlates with elevated glucose levels. The Accu-Chek SmartGuide CGM solution aims to address those critical unmet needs. Every five minutes, the Accu-Chek SmartGuide CGM sensor sends glucose values measured in real time to the Accu-Chek SmartGuide app. The Accu-Chek SmartGuide Predict app then utilises those glucose values and other available information to detect glucose patterns and predict future glucose levels. Its integrated AI-enabled predictive algorithms indicate hypoglycaemia risk within the next 30 minutes, continuously forecast how glucose levels will develop within the next two hours, and estimate the risk of nocturnal hypoglycaemia. As such, Roche’s new CGM solution is designed to alleviate people living with diabetes’ and caregivers’ concerns about nighttime hypoglycaemia and lower its risk. It aims to support informed therapy self-management decisions, enabling proactive intervention before glucose levels require immediate attention. #CEmark #CGMsystem #AccuChekSmartGuide  #AccuChek #RocheDiagnostics #Roche 

713

6 Comments

Have you ever heard of digital patient twins? I find the concept so intriguing for the future of potentially faster, safer, and more accurate clinical research. And Rene Fischer of Fraunhofer IESE does an incredible job of explaining its current state and potential. #clinicaltrials #clinicalresearch

13

1 Comment

Reflecting on recent tradeshows, it's been eye-opening to witness administrators' challenges firsthand, especially in patient communication and tracking during visits. The resounding feedback? Qure4u’s Patient Navigator and Patient Tracker features provide the perfect solution, evaluating patient experiences and streamlining staff workflows. #TheQureUNeed #HealthcareIT #DigitalHealth #PatientTracking #PatientCommunication

4

I’ve helped dozens of startups accelerate their sales in the USA. One of them is Pearl Technology AG, the leading manufacturer of patient positioning solutions in radiology. Thomas Mueller, the General Manager, came to us for help to boost their US sales. His situation was typical of many of our clients: Opportunistic selling and passive order filling from US customers get them so far, but it just doesn't do the giant US market justice. At the same time, the market is so big that it is risky to try and “wing it.” Pearl Technology needed systematic customer discovery and validation of each step of a go-to-market plan to go to the next level. This is exactly what a team from Pearl Technology and USA Launching Pad did. Here are 4 unique insights they discovered in addition to thousands of new contacts, hundreds of new LinkedIn connections, and 19 deep discovery interviews with professionals in their industry: 1. Validated the features most valued by American professionals when comparing and purchasing products. This information is now used to craft marketing messages that resonate in the USA. 2. Identified the sources where US decision-makers search for information before purchasing products. I'm glad we did the methodical work! The most effective channels were not the ones the team expected. 3. Identified all stakeholders involved in choosing and purchasing products, including what specific information each one needs. This also had some surprises. 4. Pinpointed the key sales, fulfillment, and after-sales channels that customers find most convenient: The team knows exactly where and how their products must be available. Congratulations to the Pearl Technology team for their unique go-to-market plan. I know it will be successful because it was exclusively created with feedback from actual customers. P.S: You can download the Pearl Technology case study on our website at https://lnkd.in/gXKj9VTz USA Launching Pad 🚀

13

4 Comments

The smallest things make the biggest difference. When it comes to Echocardiograms, there are very narrow margins for error. ViewPoint™ 6 with EchoPAC™ Connect is a comprehensive ultrasound reporting and image management solution with state-of-the-art cardiovascular quantification capabilities that help ensure accuracy and data consistency, boosting diagnostic confidence. Book a meeting today! #CardiacUltrasound #ViewPoint #EchoPACConnect #GEHealthCare #UltrasoundReporting

12

You are invited to our Artificial Intelligence Webinar. Register at: https://lnkd.in/gqWqqksZ In this ExecHQ Entrepreneurial Solutions (ESS) webinar series entitled: Maximizing AI for Business Growth: Strategies, Success Stories, and Solutions learn the following: Session 1: How to Convert AI Promise into ROI for Your Business Description: Learn strategies to bridge the gap between AI potential and financial returns. This webinar will cover how to think about potential AI use cases, including aligning AI initiatives with business goals, and optimizing AI investments for maximum ROI. Session 2: AI Case Studies: Thinking About AI Use Cases Across Various Parts of Your Business Description: Explore real-world examples of successful AI deployments across different business functions. This session will provide insights into how AI can enhance marketing, sales, operations, customer service, and more, using case studies to illustrate best practices and innovative approaches. Session 3: Overcoming Internal Challenges: Addressing Pain Points to Kickstart AI Implementation Description: Understand the common internal issues and pain points that businesses face when starting AI projects. This webinar will discuss strategies for overcoming resistance to change, addressing data quality and integration challenges, and ensuring executive buy-in and support for AI initiatives.

18

2 Comments

With our EHR templates you can modify and customize them to fit your specialty, workflow, and preferences. Get a faster, simpler way to capture patient data—which means you can spend more time treating patients and less time charting. Learn how to: ✍️Create a customized note for your practice 🖥️See our entire library of over 1,300 pre-made templates 🗒️See how our templates are sorted by keywords, specialty and template type https://lnkd.in/g4d3Dkgm

1

deepc closes another round of funding to accelerate its growth and solidify its role in AI deployment in the clinic !👇 Reflecting on early conversations with Dr. Franz MJ Pfister, MD, MBA, I am amazed by how his vision and yesterday's ideas have transformed into a leading radiology AI company with customers in over 30 countries ! Platforms like deepcOS offer a turnkey solution for hospitals and clinicians, easing administrative and technical burdens of AI integration. Best of luck with the next milestones, Franz and team ! 🚀 Read more on Marty Stempniak article ; link in the comment 👇 #radiology #AI #startup #funding Sofinnova Partners Bertelsmann Investments Digital Partners

21

3 Comments

If you work at a biopharma or device company and have a role in clinical trials and need to monitor cardiac safety, this article and summary can have a direct impact on the work you do every day and the patients we serve. This study demonstrated the following: ·     Patients from 18-89 can successfully take their ECG with the KardiaMobile personal 6L device. ·     Average room utilization time for the 12L group and 6L groups were 10.33±2.20 and 7.27±1.93min, respectively (p<.001), with an average time difference of 3.07min between both groups, representing a 30% reduction in ECG acquisition time. ·     If you have just 10 patients come to the office for their ECGs, you just saved 30 mins. If the patients used their 6L device from home, you would save nearly 2 hours, and this is just for 10 patients. Imagine the patient and office burden you could eliminate from your clinical trials? ·     “…The MAs performing ECGs in our EP clinic are highly experienced and may not reflect the time needed to perform a 12L ECG in other settings.” Which means it is highly likely that you will save a lot more staff and patient time than this study demonstrated at a major academic medical center. #clinicaltrials #DCT #digitalhealth #biopharma #ECG

12

Explore innovative paths in healthcare with the #JAMIA article, "Patient-centered clinical decision support: Challenges and opportunities identified from workflow execution models." This research examines the essential aspects of integrating clinical decision support systems within patient-care frameworks. By leveraging insights from workflow execution models, the authors highlight both the obstacles and the potential for these systems to enhance the patient experience and improve the decision-making process for healthcare providers. https://hubs.ly/Q02DHq5z0

15

Bringing a new medical device to market isn't easy. Here's FIVE tips to help smooth the journey to regulatory approval: 1️⃣ Perform a well-designed SOTA literature review early. This will help you understand how effectively similar devices solve the same problem, informing device design. It will also guide you on datapoints for any clinical studies, saving you time and money in later stages. 2️⃣ Apply ISO 14971:2019 risk management EARLY. This will force you to consider product design from all angles and identify risks you wouldn't otherwise know about. Effective risk mitigation often involves changes in design - easy to do early, a nightmare to do on a mature product. 3️⃣ Don't soften the design just to secure an 'easier' route to conformity. If the desired product features you have identified require a higher classification, so be it. Consider a more challenging conformity route to be a form of insurance against 'me too' competitors. 4️⃣ Take advice on clinical evidence generation and planning for clinical evaluation. Ensure that any clinical studies offer leverage, producing data that supports both regulatory approval AND future marketing activities. 5️⃣ Plan, budget, and seek advice. The worst mistake is to consider regulatory approval to be a problem for tomorrow. This is a pathway to failure. Ensure you get the right advice to help you stay ahead. Embracing and budgeting for the approval pathway will: - Get you organised - Establish a list of priorities - Impress investors, showing a clear path to market - Avoid delays and loss of commercial opportunities - Build a 'moat' of protection against competitors - Save in additional fees for unsuccessful submissions I believe that taking the right approach to medical device approval can convert the burden of regulatory compliance into a real asset for your business, with embedded leverage being the 'secret sauce'. #mantrasystems #medicaldevice #marketaccess #mdr #ivdr #eumdr

30

2 Comments

These are always important points in medtech/dental. What trends are emerging in the types of devices receiving Breakthrough designations? How can companies leverage this program to expedite the development of innovative solutions? How can companies strengthen their quality management systems to minimize the risk of recalls? What areas of non-compliance were highlighted in the latest Warning Letters? How can companies proactively address these areas to avoid regulatory scrutiny? https://lnkd.in/ggbuK_7E #FDA #MedicalDevices #Regulation #MedTech #Healthcare

8

Dear Cardiologist,   Join our partner GE HealthCare for Season 3 of their “Cardiology Coffee Breaks: The Path to Connected Care!” In just 15 minutes, unlock insights into how enterprise imaging, analytics, workflow optimizations, and more can revolutionize efficiency in your cardiology department.   Mark your calendars for the upcoming episode "Patient Work, Not Paperwork - Digitize Your Department KPI's" airing on Wednesday, May 15th. Reserve your spot today: https://lnkd.in/eEGQWEfd   Can't wait until then? Dive into the first episode, "The Road to Enterprise Imaging in Cardiology," available on demand, via the link above. Discover how enterprise imaging connects care, streamlines workflows, and saves invaluable time and resources for clinicians.   Don't miss out on this opportunity to be at the forefront of the conversation shaping the future of cardiology IT. Secure your spot now! Sign up today!   #CardiologyCoffeeBreaks #ConnectedCare #EnterpriseImaging #HealthcareInnovation #Duomed

24

NextGen Healthcare Expands Ambient Assist AI Solution for Small Practices There are a lot of exciting things happening with EHRs these days...and it is all thanks to AI. EHR vendors are adding AI capabilities at a rapid rate, but more importantly adding it in ways that make the lives of clinicians and patients BETTER. When I heard that NextGen had expanded their AI solution, I wanted to learn more. Srinivas Velamoor, NextGen's President & COO, agreed to come on camera to answer a few questions about their AI-driven ambient listening technology (Ambient Assist) and tailored solution for small practices. Check out our short-and-packed converation. https://lnkd.in/gTMqMrAa #digitalhealth #healthcare #healthIT #HITsm #healthcareAI

11

2 Comments

Discover a groundbreaking application of AI in radiology that significantly reduces radiologists' workload while ensuring high safety standards !👇 While AI is often seen as a tool to better identify and characterize abnormalities in medical imaging, its potential to alleviate our workload by identifying normal exams is profound, as demonstrated by this recent study. The research team assessed the performance of #Lunit chest radiograph AI software, which identifies 10 abnormalities, to identify normal exams, thereby potentially reducing the radiologist's burden by eliminating these normal cases from their worklist. Key findings include (quotes): - The AI system achieved an AUC of 0.92 for detecting normal chest radiographs. - 53% of normal chest radiographs were identified with a 98% NPV for urgent findings. - AI can reduce the chest radiography reporting workload by 15%. In routine practice, many radiology exams are normal yet require the same meticulous attention as abnormal ones. An AI solution that reliably identifies normal exams can prioritize our focus on more critical cases, and eventually removes normal cases from our workload. In the future, AI could potentially manage normal exams independently, presenting only the doubtful or abnormal ones to radiologists. This workflow is especially beneficial for high-volume exams with large backlogs like chest radiographs. However, for full adoption, it’s commonly believed that the AI’s specificity and NPV must approach 100% to avoid missing critical abnormalities. I agree, but only if the current standard of care and workflow can capture all abnormalities 24/7, which isn’t the case. Therefore, it’s essential to compare the AI pathway with real-world standards of care. If AI proves safer, it warrants serious consideration ! Moreover and in real life, most routine radiographs are not immediately reviewed by radiologists but by non-radiologist physicians at the point of care. Evaluating AI's impact in this context is crucial for fully realizing its potential. Importantly, the AI system in this study was not specifically trained to detect normal exams but to identify the ten most common chest abnormalities. This limitation suggests that with further development toward the detection of normality, AI systems could become even more effective at safely reducing radiologists' workload by eliminating normal exams from their worklist. Kudos to the research team for exploring this vital AI use case in radiology. I eagerly anticipate further advancements in this field! Link to the article in the comments 👇 #Radiology #AI #Normal #Workload Lunit Cancer Screening Universitair Medisch Centrum Utrecht Bram van Ginneken Radboudumc Mathias Prokop Steven Schalekamp Kicky van Leeuwen

85

6 Comments