Sie sind auf Seite 1von 2

Proposal for Research: Artificial Neural Network versus Traditional Rule-Based Approach

on Development and Maintenance of 3D Printed Organs


Introduction
When any product is being made, it is important to test for any defects with the product. This is
true whether there be design flaws, accidental manufacturing mistakes, or safety concerns.
Numerous regulations on design and safety measures are in place, but a fault detection system is
still needed for any manufacturing mistakes, which are bound to happen considering the quantity
being made for this capitalist economy. 3D printing has already started to appear in areas of
basic manufacturing and will only continue to expand as time goes on. This expansion includes
printing of organs. There are currently over 100,000 people on the waiting list, and this number
will only increase as longevity increases. As with any product, these 3D organs will need a
system to detect any flaws/malfunctions.
Fault Detection and Identification has long been the defining factor in quality control among so
many of our commercial products. As 3D printing expands its horizons from not just plastic
guns and cellphone cases, but to items such as organ printing, a more intricate defect detecting
system is necessary. Detecting problems in organs being made will save the company millions in
lawsuits in the long run, as well as keeping customers alive and well. The process for artificial
neural networks and rule-based approaches are as follows:
Traditional Rule-Based Approach
1. Accepts Inputs: Accepts a predetermined number and type of input.
2. Process: Goes through a series of rules. If the item accommodates the rule, it is
passed on the to next process or rule. If the item does not accommodate the rule, that rule
is marked as fail and the process is either halted at the error or halted after all the rules
were checked.
3. Report: Errors are reported and item is declared to have either passed or failed. If
the item has failed it is pulled for analysis.
Artificial Neural Networks
1. Accepts Inputs: Accepts a large number of inputs. These inputs can be unknown
and of various types.
2. Process: Uses characteristics of inputs given to determine whether or not there is
an error. Detects patterns without instruction.
3. Report: Errors are reported. Item is declared to have either passed or failed. If
the item has failed then it is pulled for analysis.
Current Situation
While prototypes of organs made with real human cells are already underway, larger quantities of
organs will not be produced until current prototypes are perfected. 3D organ development will
be available in the near future, and since these are manufactured items that will be keeping the

patient alive, it should be dealt with the same standards as pacemakers and similar medical
equipment.
Currently there arent any manufacturing guidelines for organs, but considering their function
some sort of quality checks and error detection should come standard. For common commercial
products, most products undergo manufacturing with rule-based error prevention. Rule-based
errors check for specific attributes and determine whether the product passes these requirements
or not. Rule based error management can not detect errors it was not programmed for.
Considering the complexity of organs, even on the microscopic level, errors that were not
originally intended or thought about could very well appear and go undetected, causing problems
for the patient. New methods of error detection are necessary for the advancement of 3D organs.
Focus
We will be researching how both artificial neural networks and rule-based methods of error
detection collect their results and what kind of errors each one misses out on. The two methods
will be evaluated based on two main categories:
1. Efficiency
Efficiency will be measured by the following categories: error detected and validity of the error,
time taken to detect errors, and how vital errors found are to organ function.
2. Cost
The cost of both systems will be taken into consideration while choosing an error detection
system. The sole reason that this focus was not listed first is because the goal of finding an error
prevention system is to choose the most reliable one to help eliminate excess cost to the company
from lawsuits. If there is only a slight variation in cost or the cost of one provides more value
over the other, the most accurate system will be chosen.
Sources
Aram, Zainab. Using Chaotic Artificial Neural Networks to Model Memory in the Brain.
Buche, C. Simulation within Simulation for Agent Decision-Making: Theoretical Foundations
from Cognitive Science to Operational Computer Model. Authors:
Fernando, Heshan. An Unsupervised Artificial Neural Network versus a Rule-Based Approach
for Fault Detection and Identification in an Automated Assembly Machine.
Jiang, Xiaomo. Fuzzy Stochastic Neural Network Model for Structural System Identification.
Milai, Ljubia. Application of Artificial Neural Network with Extreme Learning Machine for
Economic Growth Estimation.
Conclusion
As technology advances in 3D printing, our need to detect functionality flaws in organs will
increase, meaning the most precise way to detect these mistakes is key. This research will be
invaluable to the company because these findings will help promote the health and wellbeing of
the patient as well as avoiding as many unnecessary lawsuits as possible.

Das könnte Ihnen auch gefallen