Software Description
MVC app: A smartphone application for performing chemometric methods

https://doi.org/10.1016/j.chemolab.2015.08.010Get rights and content

Highlights

  • Development of MVC app as a new application for performing chemometric methods

  • Multivariate calibration methods of MLR, PCR and PLS are included in MVC app.

  • MVC app contains various options for preprocessing, rank determination and plotting.

  • Possibility of performing these options just by touching screen with no complexity

Abstract

In this work, a novel smartphone application entitled “MVC app” is developed to perform different multivariate calibration methods. This app is designed for chemists who are not expert in programming or in advanced statistics. The developed application can use any Android-powered device as an environment for running. It is an easy to use app which can simply install in your smartphone and play. Different multivariate calibration methods, such as multiple linear regression (MLR), principal component regression (PCR) and partial least squares (PLS) are included in this app. As an instance, for performing PLS modeling, first calibration and validation data sets are imported (via USB or Wi-Fi). Then, the number of latent variables (LVs) is chosen using leave-one-out cross-validation (LOO-CV). Afterwards, PLS model is built and the user can review the modeling results. In this regard, figures of merit (FOMs) of models, such as root-mean square error of prediction (RMSEP), standard error of prediction (SEP), bias and relative error (RE) and other parameters can be viewed for each analyte. Furthermore, various plotting options are included for each model. All of these options are available just by touching the screen, with no complexity that almost every chemist can use.

Introduction

Calibration is an important topic in analytical chemistry which its purpose is determination of the amount of analyte(s) of interest in unknown samples in the presence of interferences. Moreover, as a side goal, it is sometimes convenient to obtain informative information about the identity of the analytes and the interferences, if present.

Owing to the fact that analytical chemistry can involve samples that are far from simple and often contain many components, therefore, multivariate calibration (MVC) methods have been proposed in recent decades to overcome fundamental mathematical challenges occurred during analysis of these complex mixtures [1]. The main goal of MVC is development of a quantitative model for prediction and interpretation of the properties of interest (e.g., concentration) using a number of predictor variables. Among different MVC algorithms, multiple linear regression (MLR) [2], principal component regression (PCR) [2] and partial least squares regression (PLSR) [3] has attracted great attention in chemistry in recent years due to their unique properties in the solutions and wide variety of applications. However, one of the most important features of chemometric methods is development of user-friendly software in order to develop the range of users of these methods. In this regard, development of software for chemists who are not expert in programming or in advanced statistics will be an important mission for chemometricians. Great efforts have been done by different chemometric groups in this direction and different GUIs have been developed for performing MVC methods in user-friendly environments, such as PLS-toolbox [4], TOMCAT [5], MVC1 [6] and some other ones. On the other side, nowadays, smartphones are used as portable personal computers (PCs), changing our lives and how we interact to technology. The Android platform provides a vast range of applications that can be downloaded directly onto the phone. These mobile applications or “apps”, have a wide range of functionalities that can be used in many different disciplines [7]. They are easy-to-use, low price, flexible and finally anything is accessible just by touching the screen. Therefore, designing new applications for performing different types of chemometric methods can be considered as a new and interesting idea. In analytical chemistry some apps have been developed which are mostly used as transducers for colorimetric measurements by smartphone camera [8], [9], [10]. To the best of our knowledge, there is no Android-based application to perform chemometric methods. On the other side, these types of apps can be coupled to portable instruments which are highly useful for in situ and field studies. In these cases, data can be collected with instruments and then, data can be transmitted to a smartphone device through a Wi-Fi or USB connection and finally, they can be easily manipulated and processed with no extra effort.

Therefore, the aim of this work is development of a novel smartphone application (app) to perform different types of chemometric methods including multivariate calibration ones. The developed app is based on Java language programming and it can be run on any hardware supported Java, here Android [11]. This app is called “MVC” (Multivariate Calibration) and it is designed for chemists who are not expert in programming or in advanced statistics and seek user-friendly tools for multivariate calibration. In MVC app, different MVC methods, such as MLR, PCR and PLSR are included. Also, several important tools for data preprocessing (e.g., mean-centring and auto-scaling), rank determination and model evaluation (e.g., Cross-Validation (CV) [12]) and plotting and sharing facilities are included. Running MVC app does not require a serious experience; however, a basic knowledge of the underlying methods is helpful to successfully interpret the results. More details about theoretical concepts behind the implemented methods can be found elsewhere [1], [2], [3], [13].

With development of Smartphone apps, chemometric methods will be in your pocket and you can use them everywhere.

Section snippets

System requirements and installation

MVC requires an Android Smartphone with version 4.0 (IceCreamSandwich) or higher (the Android version can be checked from: Setting>About Phone>Android version). In order to import data and save logs as txt-file, MVC needs permission to read and write from phone external storage. This will be automatically reminded before installing MVC. MVC app can be installed by touching the app in your phone. MVC is available for free and you can access it by sending email to the corresponding author of this

Working with MVC app

Fig. 1 shows the flowchart of the MVC app to analyze different types of analytical data, such as spectroscopic, electrochemical and chromatographic ones. These steps are data import, selection of calibration model, optimization of calibration model, model validation and prediction of unknown samples and exportation of the results. These steps will be explained in the next section with a spectroscopic data set.

Conclusions

Nowadays, smartphones become a part of our lives and change how we look to life, science and technology. In this regard, many applications have been developed for different purposes which are related to information delivery. Indeed, these applications can play an important role in chemistry and especially in analytical chemistry. This would revolutionize portable analytical instrument usages. However, they should be as simple as possible that every non-expert user can use them. In this work, a

Independent testing

Prof. Alejandro C. Olivieri

Rosario Institute of Chemistry (IQUIR-CONICET), Suipacha 531, Rosario (2000), Argentina, e-mail: [email protected].

The authors have developed the Smartphone application “MVC App” which allows one to carry out various multivariate calibration methods. I have downloaded and installed the application on a cell phone and a tablet, both under Android as operating system, and found it to work as the authors describe. The developed application can implement

Conflict of interest

The authors declare that there are no conflicts of interest.

Acknowledgments

The authors would like to thank the Research Council of Sharif University of Technology (SUT) for their financial supports. Also, the authors would like to thank Ankit Srivastava for 2D plot class and EJML library developers (http://ejml.org).

References (16)

There are more references available in the full text version of this article.

Cited by (0)

View full text