There are multiple meanings of noise in the field of healthcare. Unwanted artifacts and graininess of medical imaging constitute imaging noise and can directly impact a doctor’s diagnosis. However, there is another type of noise that can lead to poor treatment outcomes. This type is “noise in decision-making”. Factors that can potentially sway healthcare providers from making the right choices are called noise.
Human-based healthcare delivery is subject to noise. Thus, the introduction of artificial intelligence in orthodontics has helped provide clear treatment guidelines. So, let’s have a look at how orthodontic AI models can reduce noise in decision-making.
Noise In Human Judgement
By nature, humans are inclined to use mental shortcuts called heuristics. According to experts, humans tend to use shortcuts and rules of thumb when making any decision. This allows quick analysis and enables us to reach a rational judgment briskly but it can compromise the decision’s authenticity.
As per a study, humans tend to offer two different solutions for the same problem at two different points in time. Small factors such as the time since your last meal or your current mood can directly impact your decision-making capability. An anxious mind or hungry stomach can potentially promote the chance variability of judgment. This unwanted variability in the judgment is noise and people from all walks of life are affected by this. Judges and doctors are more prone to succumbing to decision noise.
Impact Of Decision-Making Noise In Healthcare
Behavioral researchers point out that unclear decisions and noise can have detrimental effects on public health. Professionals are continuously working to minimize systematic errors and noise. Experts like Plous have identified two types of treatment predictions:
Clinical predictions: Predictions that are based on human judgment.
Actuarial predictions: Predictions based on a provided set of variables and factors.
He concluded that decisions and predictions are usually more accurate when a human does not make them. He even went on to say that decisions can be noisy/inaccurate even when he has complete access to the actuarial information. Many suggest noise auditing is an effective way but the best method of correction in this evolving world is the use of artificial intelligence.
How Can AI Address Noise In Orthodontic AI Models?
Clinical Decision Support Systems (CDSSs)
Specialized computer programs i.e., clinical decision support systems are being used in orthodontics. These systems provide expert support to the healthcare providers in diagnosis, treatment, and even prevention of diseases. The systems have shown promising results because they are based on mathematical tools.
The implementation of CDSSs in orthodontics has proven to be a great success. According to reports, clinical decision support systems reduce subjectivity, decrease diagnostic errors (noise), and increase the efficiency of diagnosis/treatment planning. The guidance system of artificial intelligence is reliable and free of noise because it performs the following functions:
Unsupervised Learning
Machine learning is the core of artificial intelligence. As the advanced systems execute a function based on the provided data, the first step in keeping noise at bay is to design ideal algorithms. In supervised learning, humans dictate which data is to be fed to the computer and what is considered “important”. These decisions (of humans) are subject to noise and can lead to the development of “model’s blind spots”.
Thus, an unsupervised learning approach in preparing algorithms is the best way to go. The algorithms train and operate without human-validated data. This helps the model uncover new anatomical/physiological relationships that are not known by the physicians. Studies show that Non-knowledge-based (unsupervised) CDSSs acquire data directly from the user (via smartphone apps, devices, etc.) and make decisions based on statistical patterns. The artificial neural networks continually learn and discover patterns within the data.
Takes Heuristics Out Of The Picture
AI and machine learning have shown great results in minimizing the disparities in decision-making. The clinical decision support systems are optimized to maximize patient health and minimize errors.
Artificial and convolutional neural networks are frequently used in AI models for healthcare. Though these network models mimic the human brain, the algorithms do not fall prey to gut feelings and intuition biases. Unlike humans, AI is not distracted by external factors. Heuristic reasoning can pose to be a hindrance in orthodontic judgments and decision-making.
Experts now believe that computer-aided heuristics can lead to better diagnosis and treatment outcomes. A high degree of uncertainty in decisions is seen in orthodontic decisions, thanks to judgment errors and cognitive biases. Research shows that the development of a proper framework for cephalometric values and computational reasoning can significantly improve orthodontists’ outcomes (especially in class Ⅲ cases).
Provides Judgement Based On Analysis
The artificial intelligence algorithms operate by analyzing the situation based on the saved data and the learned algorithm. All AI algorithms become usable only after they have been exposed to loads of data. In the case of orthodontic diagnosis and treatment planning, numerous orthodontic images (cephalometric X-rays, CBCTs, etc.) are shown to the program so it can learn and recognize patterns.
This information is paired up with decisions (that have proven to be successful) such as extraction vs. non-extraction. So, unlike human judgment, AI algorithms provide options based on the learning and continue to give the same solutions at any given time. A study showed that the diagnostic accuracy, consistency, sensitivity, and specificity of AI was better than clinicians.
Final Word
Noise in clinical decision-making is a common phenomenon in the healthcare industry. A noisy judgment by the orthodontist can impact diagnosis and treatment outcomes. Noise in decision-making is attributed to heuristic reasoning. Humans use mental shortcuts (such as heuristics) to quickly reach a diagnosis. However, this practice is prone to noise and bias. Thus, AI-based clinician decision support systems (CDSSs) are now implemented to reduce human errors and noise.
CDSSs are quick, efficient, and free of decision noise. AI achieves this feat in multiple ways. The AI algorithms are prepared using unsupervised learning. This allows the computer to learn from provided examples rather than rely on the datasets provided by a human expert.
Studies show that heuristics increase decision noise. By removing cognitive bias and heuristics, AI-based models’ decisions are free of noise. Moreover, as artificial intelligence algorithms provide answers based on datasets and patterns, you get the same solution for a problem every time.
References
- https://www.linkedin.com/pulse/decision-making-daniel-kahneman-laxmi-abhay-yesff/
- Belle, N., Cantarelli, P., & Wang, S. Y. (2024). The management of bias and noise in public sector decision-making: experimental evidence from healthcare. Public Management Review, 1-24.
- Plous, S. (1993). The psychology of judgment and decision making. Mcgraw-Hill Book Company.
- Asiri, S. N., Tadlock, L. P., Schneiderman, E., & Buschang, P. H. (2020). Applications of artificial intelligence and machine learning in orthodontics. APOS Trends Orthod, 10(1), 17-24.
- Slussareff, M. (2022). O’Neil, Cathy. 2016. Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. Crown.
- Zikos, D., & DeLellis, N. (2018). CDSS-RM: a clinical decision support system reference model. BMC medical research methodology, 18(1), 137.
- Hicks, E. P., & Kluemper, G. T. (2011). Heuristic reasoning and cognitive biases: Are they hindrances to judgments and decision making in orthodontics?. American journal of orthodontics and dentofacial orthopedics, 139(3), 297-304.
- Auconi, P., McNamara Jr, J. A., & Franchi, L. (2020). Computer-aided heuristics in orthodontics. American Journal of Orthodontics and Dentofacial Orthopedics, 158(6), 856-867.
- Shi, F., Wang, J., Shi, J., Wu, Z., Wang, Q., Tang, Z., … & Shen, D. (2020). Review of artificial intelligence techniques in imaging data acquisition, segmentation, and diagnosis for COVID-19. IEEE reviews in biomedical engineering, 14, 4-15.
- Shen, J., Zhang, C. J., Jiang, B., Chen, J., Song, J., Liu, Z., … & Ming, W. K. (2019). Artificial intelligence versus clinicians in disease diagnosis: systematic review. JMIR medical informatics, 7(3), e10010.