Bioelectronic tongue dedicated to the analysis of milk using enzymes linked to carboxylated-PVC membranes modified with gold nanoparticles

Jou nal P e-p oo Bioelec onic ongue dedica ed o he analysis o milk using enzymes linked o ca boxyla ed-PVC memb anes modi ied wi h gold nanopa icles Cla a Pe ez-Gonzalez, Co al Sal o-Comino, Fe nando Ma in-Ped osa, C is ina Ga cía-Cabezón, Ma ía Luz Rod íguez-Méndez PII: S0956-7135(22)00618-1 DOI: h ps://doi.o g/10.1016/j. oodcon .2022.109425 Re e ence: JFCO 109425 To appea in: Food Con ol Recei ed Da e: 27 Ap il 2022 Re ised Da e: 28 July 2022 Accep ed Da e: 2 Oc obe 2022 Please ci e his a icle as: Pe ez-Gonzalez C., Sal o-Comino C., Ma in-Ped osa F., Ga cía-Cabezón C. & Rod íguez-Méndez Ma í.Luz., Bioelec onic ongue dedica ed o he analysis o milk using enzymes linked o ca boxyla ed-PVC memb anes modi ied wi h gold nanopa icles, Food Con ol (2022), doi: h ps://doi.o g/10.1016/j. oodcon .2022.109425. This is a PDF ile o an a icle ha has unde gone enhancemen s a e accep ance, such as he addi ion o a co e page and me ada a, and o ma ing o eadabili y, bu i is no ye he de ini i e e sion o eco d. This e sion will unde go addi ional copyedi ing, ypese ing and e iew be o e i is published in i s inal o m, bu we a e p o iding his e sion o gi e ea ly isibili y o he a icle. Please no e ha , du ing he p oduc ion p ocess, e o s may be disco e ed which could a ec he con en , and all legal disclaime s ha apply o he jou nal pe ain. © 2022 Published by Else ie L d. Au ho s con ibu ion MR-M, CG-C, and FM-P concep ualized he idea and supe ised he wo k. CP-G and CS-C pe o med he expe imen , cu a ed he da a, and w o e he o iginal d a . FM-P in ol ed in so wa e design and de elopmen . CP-G and CS-C in ol ed in o mal analysis. CG-C and MR-M acqui ed he unding. CP-G, CS-C, FM-P, MR-M and CG-C e iewed and edi ed he pape . All au ho s p o ided eedback. Biog aphies Cla a Pe ez-Gonzalez ob ained he Ms in Nanoscience in 2019 (U. Valladolid. Spain). She is cu en ly wo king on he PhD Thesis which is dedica ed o he de elopmen o elec ochemical senso s o he analysis o oods. She is au ho o 5 scien i ic pape s. Co al Sal o-Comino ob ained he Ms in Analy ical Chemis y in 2015 (U. Complu ense. Mad id. Spain). She is cu en ly wo king on he PhD Thesis which is dedica ed o he de elopmen o elec ochemical senso s o he analysis o oods. She is au ho o 13 scien i ic pape s. Fe nando Ma in-Ped osa is ull p o esso a he Uni e si y o Valladolid and Head o he Depa men o Ma e ials Science. His esea ch is dedica ed o elec ochemis y s udies o di e en solid ma e ials. He is au ho o mo e han 80 pape s. C is ina Ga cia Cabezón, is assis an p o esso a he Enginee s school o he Uni e si y o Valladolid. She is an expe in elec ochemis y and impedance spec oscopy. She is au ho o coau ho o mo e han 50 pape s in he ield. Ma ia Luz Rod iguez-Mendez is Full p o esso o Ino ganic Chemis y a he Enginee s School o he Uni e si y o Valladolid and Head o he g oup o senso s UVASens. She is leading se e al unded P ojec s de o ed o he de elopmen o a ays o ol amme ic nanos uc u ed senso s and biosenso s o he cha ac e iza ion o oods. She is au ho o co-au ho o o e 165 publica ions (H index 44), se en books and h ee pa en s in he ield. Jou nal P e-p oo 1 Bioelec onic ongue dedica ed o he analysis o milk using enzymes 1 linked o ca boxyla ed-PVC memb anes modi ied wi h gold 2 nanopa icles 3 4 Cla a Pe ez-Gonzalez1,2,3, Co al Sal o-Comino1,2, Fe nando Ma in-Ped osa2,3, C is ina Ga cía- 5 Cabezón2,3*, Ma ía Luz Rod íguez-Méndez1,2* 6 7 1 G oup UVASENS, Escuela de Ingenie ías Indus iales, Uni e sidad de Valladolid, Paseo del Cauce, 59, 8 Valladolid 47011, Spain. 9 2 BioecoUVA Resea ch Ins i u e, Uni e sidad de Valladolid, 47011 Valladolid, Spain 10 3 Depa men o Ma e ials Science, Uni e sidad de Valladolid, Paseo del Cauce, 59, 47011 Valladolid, Spain 11 * Co espondence: [email p o ec ed], [email p o ec ed] 12 Highligh s 13 • A po en iome ic bioET speci ically dedica ed o milk analysis was de eloped. 14 • Enzymes we e co alen ly linked o memb anes combining Ca boxila ed-PVC and 15 AuNPs. 16 • The e ec i e enzyma ic immobiliza ion helped o e ain he enzyma ic ac i i y. 17 • Using SVM and ensemble me hods, nine physicochemical pa ame e s can be 18 de e mined simul aneously. 19 20 Abs ac 21 Bioelec onic ongues (bioET) made o senso s combining enzymes and nanoma e ials 22 ha e been shown o be ad an ageous due o he speci ici y o e ed by he biosenso s and 23 he enhanced sensi i i y p o ided by he nanoma e ials. In his wo k, an inno a i e bioET 24 o milk analysis is de eloped using po en iome ic biosenso s based on lac ic 25 dehyd ogenase, galac ose oxidase and u ease speci ic o he de ec ion o compounds o 26 in e es in milk (lac ic acid, galac ose and u ea). The pe o mance o he biosenso s has 27 been os e ed by co alen ly immobilizing he enzymes on memb anes o ca boxyla ed 28 poly inyl chlo ide combined wi h gold nanopa icles. The design and composi ion o he 29 biosenso s con ibu es o p ese ing he enzyma ic ac i i y, allowing limi s o de ec ion 30 in he ange o 10-5 – 10-6 M wi h excellen sensi i i y and ep oducibili y ( a ia ion 31 coe icien s anged om 1 o 5.1 %). 32 The h ee biosenso s, combined in a single de ice and coupled o a pa e n ecogni ion 33 so wa e, can disc imina e e icien ly wel e classes o milk wi h di e en a con en 34 (skimmed, semi-skimmed and whole milk) and nu i ional cha ac e is ics (calcium 35 en iched, lac ose ee and olic acid-en iched). The bioET shows an excellen 36 classi ica ion capabili y wi h an accu acy o up o 99.7%. By applying Suppo Vec o 37 Machine (SVM) analysis, he BioET can pe o m he simul aneous assessmen o eigh 38 physicochemical pa ame e s (acidi y, a , p o eins, lac ose, densi y, u ea, d y ma e and 39 non a d y ma e ) wi h sa is ac o y co ela ion coe icien s and low esidual e o s. The 40 esul s a e u he imp o ed by implemen ing ensemble me hodologies. The p oposed 41 s a egy has been demons a ed o be use ul o imp o ing he pe o mance o bioETs in 42 he dai y indus y. 43 44 Keywo ds 45 Bioelec onic ongue, milk, biosenso , gold nanopa icles 46 Jou nal P e-p oo 2 1. In oduc ion 47 48 In ecen yea s, he ield o elec onic ongues (ETs) has d i en impo an basic 49 de elopmen s (Juzhong, & Jie, 2020; Aouadi e al., 2020; Rod iguez-Mendez, De Saja & 50 González-An ón, 2016; Ha e al. 2015). Much o his p og ess is ela ed o he design o 51 new senso s which inco po a e nanoma e ials ha imp o e he sensing cha ac e is ics, 52 hanks o hei high su ace o olume a io and excellen elec oca aly ic p ope ies (Li, 53 Li, Liu, & Chen, 2019; Wang & del Valle, 2021; Sob ino-G ego io, Ba alle , So o, & 54 Esc iche, 2018; Teodo o, Shimizu, Scagion, & Co ea, 2019; Ame ico da Sil a e al., 55 2019). O he ad ances a e ela ed o new app oaches o da a managemen , including mo e 56 e icien da a educ ion me hods and imp o ed pa e n ecogni ion algo i hms and 57 classi ica ion echniques (Tian, Chen, Pan, & Deng, 2013; P ie o e al., 2013). 58 The eme gence o bioelec onic ongues (bioETs) combining classical unspeci ic senso s 59 wi h biosenso s has been a b eak h ough in he ield, because hese sys ems 60 simul aneously p o ide global in o ma ion abou he sample (as in classical ETs) plus 61 in o ma ion abou speci ic compounds ob ained om he biosenso s (Wasilewski, 62 Kamysz, & Gebicki, 2020; Skladal, 2020; Ghasemi-Vamankhas i e al., 2012; Yhan e 63 al., 2021; Ha e al., 2017). The pe o mance o elec ochemical biosenso s can be u he 64 imp o ed by combining enzymes o o he biological bio ecep o s wi h nanoma e ials. 65 Nanoma e ials p o ide an e ec i e pla o m o he immobiliza ion o biomolecules, 66 inducing unique pe o mance cha ac e is ics in e ms o sensi i i y and speci ici y. Some 67 examples o ol amme ic bioETs based on combina ions o enzymes and nanoma e ials 68 ha e ecen ly been epo ed. Fo ins ance, an a ay o med by phenol oxidases and 69 glucose oxidase combined wi h nanopa icles has been success ully used o analyze 70 g apes and mus s (Ga cia-Cabezón e al., 2020; Ga cia-He nandez e al., 2019). Human 71 as e ecep o s combined wi h ca bon nano ubes (CNTs) o polypy ole nano ubes ha e 72 been used o o m a ield e ec ansis o wi h human- ongue-like selec i i y (Kim e al., 73 2011; Song e al., 2012). 74 Milk is a complex mix u e ha con ains many di e en compounds, including 75 ca bohyd a es (mainly lac ose), a s, p o eins (casein o whey), mine als (such as calcium) 76 and many o he miscellaneous cons i uen s. E- ongues ha e been de eloped and applied 77 o he dai y indus y in quali y con ol, e alua ion o as e o eshness, de ec ion o 78 adul e a ions, o igin ecogni ion, e c. (Ciosek, 2016). These p e ious wo ks ha e used 79 di e en ypes o elec odes and ma e ials (Winquis e al. 1998; Wei, Wang, & Jin, 2013; 80 Pascual e al., 2018; Yu e al., 2015; Li e al., 2015; Ciosek, & W oblewski, 2015; Tazi 81 e al., 2018; Dias e al., 2009; Pé ez-González e al., 2021; Yang e al., 2021; H uška e 82 al., 2010; Collie e al. 2003; Valen e e al., 2018; Scagiona e al., 2016). Only a ew 83 a emp s ha e been made o in oduce nanoma e ials in ETs applied o he dai y indus y. 84 They include an a ay o ol amme ic elec odes modi ied wi h nanos uc u ed Laye - 85 by-Laye ilms (Sal o-Comino e al. 2018), a po en iome ic ET using senso s modi ied 86 wi h nanopa icles (Me can e e al., 2015) and an impedime ic ET using elec ospun 87 nano ibe s (Ohlson e al., 2017). Howe e , due o he complexi y o milk, he analysis 88 using ETs is no a comple ely sol ed p oblem and new de elopmen s in he ield a e 89 equi ed. 90 The p oposal he e is o ake a s ep o wa d in he ield o bioETs by de eloping no el 91 senso s combining enzymes speci ic o compounds p esen in milk (galac ose, u ea and 92 lac ic acid) wi h nanoma e ials. Galac ose and i s con en is an impo an indica o o milk 93 quali y and i s con en can be measu ed wi h indi idual galac ose oxidase (GaOx) 94 biosenso s (Ohlson e al., 2017; Kanyong, K ampa, Aniweh, & Awanda e, 2019; Mangan 95 e al., 2018; Nguyen e al. 2016). Few examples can be ound in he li e a u e, whe e 96 Jou nal P e-p oo 3 GaOx has been combined wi h nanoma e ials such as g aphene (Çakı oğlu e al. 2019) o 97 nanopa icles (Miglio ini e al., 2018). The de ec ion o u ea is also o p ime impo ance 98 o assess he nu i ional p og am o cows and can indica e unde lying pa hological 99 p oblems. Few examples o indi idual nanobiosenso s o he de ec ion o u ea ha e been 100 epo ed. They a e based on he combina ion o u ease wi h nanopa icles (Jakha & 101 Pundi , 2018) o nano ibe s (Jia e al. 2011). Finally, he con ol o lac ic acid is essen ial 102 o e alua e he e men a ion o lac ose due o lac ic bac e ia. O e he las ew yea s, some 103 examples o biosenso s based on lac a e dehyd ogenase (LDH) combined wi h 104 nanoma e ials ha e been epo ed (Rahman e al. 2009). 105 In enzyme-based biosenso s, he use o an adequa e me hod o immobilize he enzymes 106 is c ucial o p ese e he enzyma ic ac i i y and a oid leakages (Nguyen, & Kim, 2017). 107 Co alen immobiliza ion has he ad an age o high su ace loading and low p o ein loss 108 (Zucca, & Sanjus , 2014; Lee e al., 2017). Ou p oposal he e is o de elop an 109 immobiliza ion memb ane using ca boxyla ed PVC (C-PVC) -ins ead o he classical 110 PVC- whe e enzymes can be co alen ly linked using a co alen eac ion be ween 111 ca boxyl g oups o he C-PVC and he amines on he p o ein. 112 In he ETs, i is also impo an o selec he bes chemome ic me hods o p ocess he da a. 113 Unsupe ised and supe ised analysis me hods, such as p incipal componen analysis 114 (PCA), linea disc imina ion analysis (LDA), suppo ec o machines (SVM) o weighed 115 k-nea es neighbo analysis (KKNN), ha e been ex ensi ely applied (Skladal, 2020). One 116 o he eme ging ends in da a analysis is he combined use o s a is ical algo i hms 117 h ough ensemble me hodologies, whe e he ou pu s o he di e en algo i hms a e 118 combined in a decision usion s a egy o c ea e a single esponse o a gi en p oblem 119 (Zhou, 2012). Howe e , his s a egy has ba ely been applied in he ield o ETs, whe e 120 hey could ep esen a g ea ad ance in complex media analysis such as milk. 121 In summa y, he aim o his wo k was o de elop a po en iome ic bioET based on 122 memb anes made o ca boxyla e PVC modi ied wi h nanopa icles. The ca boxyla e PVC 123 is used o co alen ly link he enzymes able o de ec compounds in milks: GaOx, LDH 124 and U e, which ha e been selec ed o hei abili y o de ec impo an componen s in 125 milk. Once p epa ed and cha ac e ized, he sensing uni s a e combined in a single de ice 126 o ob ain a bioET ha is used o analyze and classi y 12 classes o milk wi h di e en 127 nu i ional cha ac e is ics and o p edic he eigh physicochemical pa ame e s mos 128 commonly used in he dai y indus y o quali y con ol. In his wo k, a i s app oach o 129 an ensemble me hodology ou ine is also p oposed o he co ela ion o da a ob ained 130 wi h he bioET wi h physicochemical pa ame e s. 131 132 2. Ma e ial and me hods 133 All he eac an s we e o analy ical g ade and we e used wi hou u he pu i ica ion. They 134 we e pu chased o m Sigma-Ald ich (S .Louis, USA). All he solu ions we e p epa ed in 135 MilliQ deionized wa e (Me ck, KGaA, Da ms ad , Ge many). 136 137 2.1 Milk samples 138 A se o 120 milk samples co esponding o 12 ypes o comme cial milk ypes ( en 139 eplicas om each milk) we e included in he s udy. This se was o med by milks wi h 140 di e en a con en (skimmed, semi-skimmed and whole milk) and nu i ional con en 141 (lac ose- ee, calcium-en iched, and olic acid-en iched milk). The milks we e analyzed 142 using adi ional s anda d chemical me hods: he i a ion me hod o acidi y (ISO 143 22113:2012), he Hyd ome e me hod o densi y (ISO 2449:1974), he G a ime y Röse- 144 Jou nal P e-p oo 4 Go lieb me hod o a con en (ISO 1211:2010), he Kjeldahl me hod o p o ein con en 145 (ISO 8968-1:2014), HPLC o de e mine he lac ose con en (ISO 22662:2007), and 146 In a ed spec oscopy o he u ea con en (ISO 9622:2013). To al d y ma e (DM) and 147 non- a d y ma e (NFDM) we e also analyzed (ISO 6731:2010) (In e na ional 148 O ganiza ion Fo S anda diza ion, 2021). The physicochemical da a a e summa ized in 149 Table 1. 150 151 Table 1. Milk samples and physicochemical pa ame e s es ablished by adi ional 152 s anda d me hods 153 Sample Fa con en Nu i ional desc ip ion Acidi y (ºD) Densi y (g/ml) Fa (%m) P o eins (%m) Lac ose (%m) NFDM (%m) DM (%m) U ea (mg/ml) S1 Skimmed Classic 12.55 1031.55 0.31 3.3 5 9.02 9.33 387 S2 Skimmed Calcium 15.82 1039.47 0.29 3.93 5.59 10.51 10.8 724 S3 Skimmed Lac ose F ee 12.66 1033.57 0.32 3.29 0.36 9.02 9.33 <10 S4 Skimmed Folic Acid 12.57 1033.7 0.40 3.29 4.95 9.04 9.43 586 S5 Semi- Skimmed Classic 12.55 1031.6 1.56 3.27 4.91 8.91 10.47 355 S6 Semi- Skimmed Calcium 16.06 1037.29 1.55 3.9 5.49 10.40 11.95 597 S7 Semi- Skimmed Lac ose F ee 12.19 1032.09 1.59 3.31 0.42 8.99 10.57 <10 S8 Semi- Skimmed Folic Acid 12.95 1032.38 1.64 3.21 4.93 8.94 10.58 638 S9 Whole Classic 12.17 1029.38 3.56 3.21 4.85 8.78 12.33 388 S10 Whole Calcium 15.86 1035.71 3.55 3.91 5.54 10.45 14.0 769 S11 Whole Lac ose F ee 11.98 1029.4 3.59 3.23 0.31 8.82 12.41 <10 S12 Whole Folic Acid 12.72 1030.55 3.1 3.18 4.94 8.92 12.02 792 154 155 2.2 Senso s and biosenso s 156 Gold nanopa icles we e syn he ized by educ ion o e achlo oau ic in he p esence o 157 isodium ci a e as he educing agen , using he classical Tu ke ich me hod (Kimling e 158 al., 2006). The colloid ob ained was cha ac e ized by UV-Vis, showing a maximum a 159 537 nm. The concen a ion o AuNPs was calcula ed by Bee ’s law, wi h a pa icle 160 concen a ion esul o 5.98 x 10-11 M and a diame e o 52.1 nm (Haiss, Nguyen, 161 A eya d, & Fe nig, 2007). 162 163 Senso s we e based on polyme ic memb anes made o ca boxyla ed PVC [poly ( inyl 164 chlo ide) ca boxyla e] (C-PVC) as he polyme ic ma ix. The C-PVC was mixed wi h an 165 addi i e (oleyl alcohol) and a plas icize [(bis(1-bu ylpen yl) adipa e (named plas icize 166 A), is(2-e hylhexyl) phospha e (named plas icize B) o 2-ni ophenyl-oc yle he 167 (named plas izice C)] using e ahyd o u ane as he sol en . A second se o senso s was 168 p epa ed by in oducing gold nanopa icles in he memb ane. 169 The memb anes desc ibed abo e we e modi ied wi h galac ose oxidase (GaOx) om 170 Dac ylium dend oides (Sigma-Ald ich, S . Louis, USA), lac a e dehyd ogenase (LDH) 171 om Mus musculus (Roche diagnos ics, Indianapolis, USA), and u ease (U e) om 172 Jou nal P e-p oo 5 Cana alia ensi o mis (Sigma-Ald ich, S . Louis, USA). The enzymes we e co alen ly 173 linked o he su ace o he polyme ic memb ane using he ca bodiimide me hod 174 (Kazenwadel, Wagne , Rapp, & F anz eb, 2015). The eac ion was ca ied ou in wo 175 s eps. Fi s , he ca boxylic g oups o he C-PVC we e ac i a ed by means o EDC (1- 176 E hyl-3-(3-dime hylaminop opyl) ca bodiimide. Then, he enzyme was added and a 177 pep ide bond was o med be ween he ca boxylic g oups on he C-PVC and he supe icial 178 amino side chains o he enzyme. As a esul o he combina ion o he six memb anes 179 wi h each o he h ee enzymes (GaOx, U e and LDH) a se o 24 memb anes we e 180 ob ained (Table 2). 181 182 Table 2. Composi ion o he senso s. 183 184 Senso C-PVC (w/w%) Addi i e (w/w %) Plas icize (P) Type (w/w%) AuNPs (w/w%) Enzyme A 32 3 A 65 - - A-GaOx GaOx A-U e U e A-LDH LDH B 32 3 B 65 - - B-GaOx GaOx B-U e U e B-LDH LDH C 32 3 C 65 - - C-GaOx GaOx C-U e U e C-LDH LDH A-AuNP 32 3 A 55 10 - A-AuNP-GaOx GaOx A-AuNP-U e U e A-AuNP-LDH LDH B-AuNP 32 3 B 55 10 - B-AuNP-GaOx GaOx B-AuNP-U e U e B-AuNP-LDH LDH C-AuNP 32 3 C 55 10 - C-AuNP-GaOx GaOx C-AuNP-U e U e C-AuNP-LDH LDH 185 186 The bioET was designed using a me hac yla e ube, in which 24 holes (0.3 cm diame e ) 187 we e d illed. The holes we e hal - illed wi h an epoxy sil e esin (EPO-TEK, Bille ica, 188 USA) and he esin was co e ed wi h one o he 24 memb anes. The inne pa o he 189 sil e epoxy esin was connec ed o a da a acquisi ion sys em (Agilen Da a Acquisi ion 190 Swi ch Uni 34970A). In all measu emen s, he Ag/AgCl elec ode was used as he 191 e e ence elec ode. Figu e 1 shows he scheme o he designed bioET sys em. 192 193 Jou nal P e-p oo 6 194 195 196 Figu e 1. Scheme o he bioET designed in his wo k. A) Da a Acquisi ion Swi ch; B) 197 Re e ence elec ode; C) Elec onic ongue body; D) Enzyme co alen ly linked; E) C-PVC 198 memb ane; F) Sil e epoxy esin and coppe wi e. 199 200 The po en iome ic measu emen s we e ca ied ou by imme sing he senso a ay in a 201 100 ml glass cell con aining he s anda d solu ions o he milk samples. S anda d solu ions 202 o compounds usually ound in milk (KCl, CaCl2, galac ose, u ea and lac ic acid) we e 203 p epa ed in a phospha e bu e (0.1M, pH 7) wi h concen a ions anging om 1 × 10−4 204 o 1 × 10−2 M. The milks we e dilu ed 1:1 in phospha e bu e and measu ed wi hou 205 u he modi ica ion. In addi ion, nico inamide adenine (NAD+) (Roche diagnos ics, 206 Indianapolis, USA) was added o he s anda d solu ions in o de o simula e he le els 207 usually p esen in milk ( inal concen a ion 12 mM) (Fox, & McSweeney,1998). A e 208 imme sing he elec odes in he co esponding sample, he memb ane po en ials we e 209 egis e ed e e y h ee seconds. The signals we e s abilized a e 5 minu es (a e age 210 a ia ion o 1.6 mV/decade be ween each eading). 211 The po en ials ob ained om he senso a ay we e used as he inpu a iables o 212 mul i a ia e analysis. P incipal Componen Analysis (PCA) was used o es ima e he 213 disc imina ion abili y o he mul isenso y sys em. A Suppo Vec o Machine (SVM) 214 was applied o es ablish co ela ions wi h he physicochemical pa ame e s ob ained using 215 adi ional me hods (Theodo e, & Robin, 2006; Co es, & Vapnik, 1995). Addi ionally, 216 he SVM was applied o elabo a e classi ica ion models. Finally, an app oach owa ds 217 ensemble me hods was implemen ed by applying S ochas ic G adien Bos ing o 218 eg ession (F iedman, 2002). The s a is ical analysis was pe o med by using Ma lab 219 R2020b (The Ma hwo ks Inc., Na ick, USA), RKWa d 0.7.1, and he Ca e package 220 (Kuhn, 2008). 221 222 3. Resul s and discussion 223 224 3.1 De elopmen and op imiza ion o he senso a ay 225 In o de o ob ain e icien po en iome ic biosenso s, he immobiliza ion o he enzymes 226 on he polyme ic memb ane was accomplished using ca boxyla ed PVC (C-PVC) ins ead 227 o he ba e PVC classically used o ab ica e po en iome ic senso s (Tazi e al., 2018; 228 Dias e al., 2009). Using C-PVC, he enzymes can be immobilized by es ablishing a 229 co alen link be ween he ca boxyla e g oups o he memb ane and he amine g oups o 230 Jou nal P e-p oo 7 he enzymes. In addi ion, memb anes we e doped wi h AuNPs o u he inc ease he 231 in ensi y o he signals. As obse ed in Figu e 2, he memb ane po en ial inc eased wi h 232 he con en o AuNPs in he memb ane. Fo ins ance, he sensi i i y alues ob ained om 233 he slopes o he calib a ion cu es owa ds galac ose we e 17.23 mV o senso A 234 (wi hou AuNPs), 19.13 mV o senso A-Au con aining 5% o AuNPs, and 32.22 mV 235 o senso A-Au con aining 10% o AuNPs. Highe concen a ions o AuNPs did no 236 p oduce any u he imp o emen in he sensi i i y alues. Based on hese indings, he 237 decision was aken o se he AuNPs con en a 10%. 238 239 240 241 242 Figu e 2. Response o senso A (wi hou AuNPs), A-AuNPs5% and A-AuNPs10% o 243 inc easing concen a ions o galac ose. 244 245 Once he composi ion o he memb anes had been op imized, he enzymes GaOx, U e 246 and LDH we e immobilized a he memb ane su ace and he esponses o he ob ained 247 biosenso s we e analyzed. As obse ed in Figu e 3, he in ensi y o he esponses 248 p oduced by a ba e C-PVC memb ane we e lowe han hose ob ained when he enzymes 249 we e co alen ly linked o he memb ane. Taking he case o u ea as an example, he 250 measu ed ol age inc eased om 0.057 V in he ba e C-PVC senso (senso A) o 0.119 251 V in he AuNP modi ied senso (A-AuNP). The enzyme addi ion inc eased he in ensi y 252 o he esponses (0.156 V in A-U e); and hey inc eased e en u he o 0.276 V in A- 253 AuNP-U e when he enzyme was combined wi h AuNPs. Simila esul s we e ob ained 254 o LDH o GaOx. 255 These esul s indica e ha he enzymes a e p ope ly immobilized a he su ace o he 256 memb ane and he enzyma ic ac i i y is e ained. The syne gis ic e ec obse ed when 257 C-PVC and AuNPs a e combined in he suppo memb ane is also wo h no ing. 258 259 Jou nal P e-p oo 14 wi hin each g oup o milks acco ding o i s nu i ional con en . The e o e, by 402 inco po a ing enzymes and AuNPs, he disc imina ion capaci y o he sys em has been 403 inc eased. 404 405 3.3 Analysis o milk wi h he bioET: Classi ica ion models 406 The milk classi ica ion analysis was based on he ea u es om he nine senso s ha make 407 up he simpli ied bioET by applying he Suppo Vec o Machine classi ica ion me hod 408 (SVMC). The Suppo Vec o Machine (SVM) is a ke nel-based supe ised pa e n 409 ecogni ion echnique, es ablished by Co es and Vapnik and based on s a is ical lea ning 410 heo y (Co es, & Vapnik, 1995). Compa ed wi h o he app oaches, SVM possesses he 411 ad an ages o a oiding o e i ing, is capable o es ablishing non-linea co ela ions 412 be ween da a se s and o dealing wi h high-dimensional inpu . 413 The SVM classi ica ion chosen was based on he adial basis unc ion (RBF) as a 414 nonlinea ke nel app oxima ion, de ined as 415 416 𝑲(𝒙𝒊− 𝒙𝒋) = 𝒆𝒙𝒑⁡((−𝜸 ∥ 𝒙𝒊− 𝒙𝒋∥𝟐),⁡⁡⁡⁡⁡⁡𝜸 > 0 417 whe e xi and xj a e he aining ec o s o he inpu da a, and γ is he ke nel pa ame e . 418 Be o e he alida ion s age, o achie e a be e pe o mance, he ke nel unc ion penal y 419 pa ame e (C) and he ke nel pa ame e γ in he SVM we e op imized. To op imize hese 420 pa ame e s, he g id sea ch me hod was applied, whe e app oaches we e made using 421 log2C and log2γ, a ying om [10, 10] a one in e al (Co es, & Vapnik, 1995). The g id 422 poin s o (C, γ) we e con i med h ough he alida ion accu acy in he [10, 10] g id. The 423 esul s showed ha he bes alida ion accu acy was achie ed when C=1 and γ=0.1. Due 424 o he ela i ely small numbe o samples a ailable, he lea e-one-ou c oss- alida ion 425 me hod was used o be e e alua e he ue success a e ha can be eached wi h he 426 SVM. 427 The classi ica ion o he samples was ca ied ou in wo s eps. Ini ially, a s udy was 428 p oposed aimed a de e mining whe he he milk samples could be classi ied based on 429 hei lac ose con en (p esence o absence o lac ose), as well as olic acid and calcium 430 con en (samples wi h o wi hou en ichmen s in calcium o olic acid). This led o he 431 de elopmen o h ee di e en classi ica ion models. 432 The esul s ob ained o each o he models we e he ollowing: 98.2% calib a ion 433 accu acy and 97.2% alida ion accu acy o milk samples wi h o wi hou lac ose; 96.3% 434 calib a ion accu acy and 95.8% alida ion accu acy o samples wi h olic acid 435 en ichmen s; and inally, 97.8% accu acy o he calib a ion and 97.1% in he alida ion 436 was achie ed in he classi ica ion model o de e mine which milk samples we e en iched 437 in calcium. All he classi ica ion models de eloped in his app oach we e able o es ablish 438 ma hema ical models wi h high accu acy alues. 439 A second app oach was aken as an a emp o classi y he analyzed milk samples 440 acco ding o hei nu i ional composi ion and hei a con en , which esul ed in a o al 441 o wel e ca ego ies. By applying SVMC, he esul s ob ained o he simpli ied bioET 442 showed 99.7% accu acy in he calib a ion and 98.4% accu acy in he alida ion. These 443 esul s de e mined ha he elec onic ongue de eloped wi h nine senso s was able o 444 classi y milk samples acco ding o hei nu i ional con en as well as o hei a con en . 445 446 3.4 P edic ion o chemical pa ame e s: Co ela ions be ween elec onic ongue and 447 chemical analysis 448 Jou nal P e-p oo 15 One o he main ad an ages o ETs is he possibili y o p edic he concen a ion o se e al 449 componen s in a single measu emen . Fo his pu pose, ma hema ical models mus be 450 de eloped o es ablish co ela ions be ween da a p o ided by he senso a ay and 451 physicochemical da a measu ed by adi ional me hods. I is expec ed ha he p esence 452 o biosenso s could help o achie e good co ela ions wi h speci ic compounds. 453 The simpli ied bioET de eloped he e was used o p edic pa ame e s commonly used o 454 assess he g oss composi ion o milks, including he o al amoun o a s, o al p o eins 455 (casein o whey), ca bohyd a es (lac ose), u ea, and o al solids (d y ma e and non- a 456 d y ma e ) which is he esidue le when wa e and gases a e emo ed. Only ew 457 a emp s o use ETs o e alua e he chemical composi ion o milk ha e been epo ed 458 p e iously (H uska e al. 2010; Sal o-Comino e al. 2018; Pé ez-González e al. 2021). 459 Suppo Vec o Machine eg ession was used o de e mine he na u e o he ela ionships 460 be ween he da a collec ed by he bioET and he physicochemical pa ame e s. To o ecas 461 acidi y, densi y, pe cen age o p o ein, lac ose, a , DM and NFDM, he Radial Basis 462 Func ion was chosen as he co e unc ion, since i can handle non-linea in e ac ions 463 be ween he senso inpu s and he a ge cha ac e is ics. The eg ession models we e 464 c ea ed using SVM Reg ession (epsilon SVM, ke nel ype: adial basis unc ion, C alue: 465 1, c oss alida ion segmen s size: 15, and s anda d de ia ion weigh ing p ocess in all 466 cases). 467 As obse ed in Table 5, he alues ob ained o he coe icien s o co ela ion and e o s 468 o he calib a ion and he p edic ion eached alues o R2 abo e 0.98 o calib a ion and 469 p edic ion, wi h low e o s (RMSE) be ween 0.101 and 0.139. These high co ela ion 470 coe icien s could be due o he speci ici y induced by he p esence o he biosenso s. In 471 ac , he biosenso con aining galac ose oxidase p o ides da a abou galac ose; LDH can 472 gi e in o ma ion on lac ic acid, which is in u n ela ed o he acidi y o he milk; while 473 u ease can accoun o he le els o u ea. The good co ela ions wi h acidi y, densi y a 474 and d y ma e can be a ibu ed o he ac ha he enzymes con ained in he a ay a e 475 sensi i e o pH. In addi ion, po en iome ic measu emen s a e sensi i e o he pe cen age 476 o wa e (which is in e sely p opo ional o densi y) and o he a con en (di ec ly ela ed 477 o he conduc i i y and he double laye a he elec ode su ace). These esul s show ha 478 he educed bioET is capable o es ablishing good co ela ions wi h he physicochemical 479 pa ame e s hanks o he selec ion o he sui able senso s in p e ious s eps o his wo k. 480 I we compa e he esul s o he eg ession wi h he p e ious wo k (Pé ez-González e al. 481 2021) we obse e an inc ease o he co ela ion coe icien s as well a educ ion in he 482 e o s (RMSE). The inc ease in R2 o lac ose and acidi y is especially ema kable. 483 Co ela ion coe icien s ha e imp o ed om alues o 0.96 and 0.90 espec i ely in he 484 alida ion, o 0.99 in bo h cases. These esul s demons a e he e ec i eness o he use o 485 biosenso s in he composi ion o a ET p o iding speci ic in o ma ion on compounds o 486 in e es in milk, such as lac ose, wi hou losing global in o ma ion o he sample. 487 488 Table 5. Co ela ion pa ame e s om he SVM eg ession analysis. 489 490 491 492 Pa ame e s Acidi y Densi y %P o eins %Fa %Lac ose %DM %NFDM U ea SVM R2C 0.9953 0.9910 0.9941 0.9956 0.9924 0.9933 0.9991 0.9915 RMSEC 0.1177 0.1376 0.1088 0.1018 0.1093 0.1233 0.1146 0.1187 R2P 0.9946 0.9903 0.9944 0.9951 0.9928 0.9927 0.9982 0.9902 RMSEP 0.1181 0.1396 0.1092 0.1055 0.1113 0.1281 0.1151 0.1193 Jou nal P e-p oo 16 3.5 Ensemble me hod de elopmen 493 Al hough he SVM was e y capable o es ablishing ma hema ical models o he co ec 494 classi ica ion o he milk samples and he p edic ion o he physicochemical pa ame e s; 495 he e, we aimed o go a s ep u he by es ablishing co ela ion models using ensemble 496 me hodologies. 497 In he con ex o machine lea ning, ensemble me hods a e commonly de ined as a 498 machine lea ning sys em, designed wi h a se o independen models wo king in pa allel, 499 whose ou pu s a e combined wi h a decision usion s a egy o c ea e a single esponse 500 o a gi en p oblem. The e o e, an ensemble me hod aims o combine se e al sepa a e 501 models o achie e a be e esul han each indi idual me hod in e ms o consis ency and 502 accu acy (Zhou, 2012). 503 The i s s ep in de eloping an ensemble me hod is o selec he indi idual me hods. 504 Di e en algo i hms may lead o di e en esul s o he same da a by imposing a speci ic 505 s uc u e o i . Mo eo e , he e is no single algo i hm able o pe o m consis en ly well 506 o di e en p oblems and he e a e no clea ules o ollow while selec ing indi idual 507 algo i hms o a gi en p oblem. 508 In p inciple, any indi idual models could be used as long as hey a e sui able o he 509 da ase . In his wo k case, he Ca e Package de eloped o R is used o selec he 510 indi idual algo i hms (Kuhn, 2008). The gene a ed models should be as di e en om 511 each o he as possible. A high le el o di e si y means ha hey will be able o cap u e 512 di e en in o ma ion abou he da a and can o e come he weaknesses o single 513 echniques, since each echnique handles he e o made by he o he s. 514 S a ing wi h he SVM eg ession model (s mRadial), i e models we e selec ed o ensu e 515 hei di e si y. Fo his, he "max.dissim" unc ion o he Ca e Package was used, in 516 which he Jacca d dissimila i y unc ion was selec ed as he di e si y c i e ion. The 517 models selec ed we e: Suppo Vec o Machine (s mRadial), Quasi- ecu en Neu al 518 Ne wo ks (q nn), Cubis Reg ession Model (cubis ), Weighed k-nea es neighbo (kknn), 519 and Bagged Ea h (bagEa h). 520 The suppo ec o machine was chosen as he s a ing model due o i s g ea pe o mance 521 in he p e ious sec ion o his wo k. Fu he mo e, SVM is a powe ul me hod widely used 522 in he de elopmen o ensemble models. 523 Once he models had been selec ed, he o iginal da a we e spli in o wo se s: a aining 524 se con aining 75% o he o iginal da a o be used in he calib a ion p ocess o each 525 algo i hm, and a es ing se co e ing he emaining 25% o he da a o alida ion. I is 526 essen ial o e i y ha bo h se s o da a a e ep esen a i e o all he ecognized ca ego ies; 527 consequen ly, he pe cen age o each se is compu ed in ela ion o he o al da a as well 528 as he amoun o da a in each ca ego y. 529 Each algo i hm was execu ed indi idually, bu he con ol pa ame e s o all o hem we e 530 es ablished be o ehand. Valida ion was pe o med using epea ed 10- old c oss- 531 alida ion, o es ablish easonable alues o he uning pa ame e s and andom sea ch 532 was es ablished as he p e e ed me hod. A e each indi idual algo i hm was applied, he 533 S ochas ic G adien Boos ing (gbm) me hod was used o gene a e he ensemble h ough 534 he Ca e Ensemble package in R (Kuhn, 2008). 535 S ochas ic G adien Boos ing is a machine lea ning algo i hm, able o pe o m 536 classi ica ion and eg ession p oblems. G adien Boos ing is especially con enien , 537 because o i s compu a ional e iciency and obus ness o o e i ing, as a simple 538 echnique o de elop ensemble decision ees by c ea ing aining ees on subsamples o 539 he aining da ase (F iedman, 2002). 540 Jou nal P e-p oo 17 Table 6 shows he alues ob ained o he co ela ion and e o coe icien s o he 541 calib a ion and p edic ion ob ained by he ensemble. The coe icien s o co ela ion and 542 mean e o s o he calib a ion and p edic ion eached alues o R2 abo e 0.9992 o bo h 543 calib a ion and p edic ion, wi h low e o s (RMSE) be ween 0.0033 and 0.0172. 544 545 Table 6: Co ela ion pa ame e s om he ensemble eg ession analysis. 546 547 Conside ing he high alues o he co ela ion pa ame e s achie ed wi h SVM eg ession, 548 i was expec ed ha he esul o he eg ession ensemble would each nea ly 100% 549 p ecision while es ablishing co ela ions, since he e is a e y educed numbe o e o s 550 in he o iginal model. Howe e , he in en ion in his sec ion is no o ensu e he capabili y 551 o he simpli ied bioET o es ablish co ela ions wi h he s udied pa ame e s, bu o 552 demons a e he possibili y o combining he de eloped sys em wi h ensemble 553 me hodologies ha could be applied in he s udy o u u e and mo e complex samples, 554 whe e he se ings may no be as good as hey should be. 555 556 557 4. Conclusions 558 In his wo k, a bioET wi h imp o ed cha ac e is ics was de eloped and used o p edic 559 he chemical cha ac e is ics o milk wi h unp eceden ed accu acy. The sys em 560 inco po a es biosenso s based on memb anes o ca boxyla ed PVC (C-PVC) con aining 561 gold nanopa icles (AuNPs), whe e GaOx, LDH and U e we e e ec i ely immobilized. 562 The de eloped biosenso s and he associa ed me hodology ha e esul ed in a bioET whe e 563 he enzymes can wo k simul aneously while also p ese ing he enzyma ic ac i i y. 564 Nanopa icles ha e p o en o ha e a po en ial o ampli y he elec ochemical signals. 565 The biosenso s ha e shown excellen sensi i i y and ep oducibili y owa ds s anda d 566 solu ions o compounds usually ound in milk (CaCl2, KCl, u ea, lac ic acid and 567 galac ose), wi h excellen sensi i i y and ep oducibili y, showing LODs o 10-6 M. 568 The bioET was success ully used o disc imina e be ween milks by applying PCA based 569 on hei nu i ional con en . The bioET shows an excellen classi ica ion capabili y and 570 can classi y milk wi h di e en composi ions by applying SVM wi h accu acies abo e 571 95%. The sys em can p edic he acidi y, densi y, %p o eins, %lac ose, % a and d y 572 ma e wi h low e o s and high co ela ion coe icien s. The esul s show ha he SVM 573 models cons uc ed wi h he e- ongue and physicochemical pa ame e s ha e po en ial o 574 use in simul aneously assessing 8 pa ame e s, hus educing he ime o analysis. 575 Mo eo e , i has been p o ed ha applying ensemble me hodologies can u he imp o e 576 he co ela ion be ween he bioET da a and he physicochemical pa ame e s. 577 In es iga ions in o he e iciency o he p o o ype de ices can c ea e new applica ion 578 possibili ies and sugges success ul implemen a ions in eal applica ions. 579 580 Decla a ion o Compe ing In e es 581 The au ho s decla e ha hey ha e no known compe ing inancial in e es s o pe sonal 582 ela ionships ha could ha e appea ed o in luence he wo k epo ed in his pape . 583 Pa ame e s Acidi y Densi y %P o eins %Fa %Lac ose %DM %NFDM U ea Ensemble R2C 0.9997 0.9999 0.9999 0.9994 0.9999 0.9999 0.9999 0.9999 RMSEC 0.0097 0.0053 0.0041 0.0164 0.0042 0.0037 0.0033 0.0040 R2P 0.9994 0.9997 0.9998 0.9992 0.9998 0.9998 0.9998 0.9998 RMSEP 0.0102 0.0075 0.0056 0.0172 0.0051 0.0045 0.0041 0.0048 Jou nal P e-p oo 18 Funding: 584 This wo k was suppo ed by MICINN-FEDER (RTI2018-097990-B-100), Conseje ía de 585 Educación JCyL- FEDER (VA202P20), EU-FEDER p og am (CLU-2019-04) and 586 «In aes uc u as Red de Cas illa y Leon (INFRARED)» 587 588 Au ho s con ibu ion 589 MR-M, CG-C, and FM-P concep ualized he idea and supe ised he wo k. CP-G and 590 CS-C pe o med he expe imen , cu a ed he da a, and w o e he o iginal d a . FM-P 591 in ol ed in so wa e design and de elopmen . CP-G and CS-C in ol ed in o mal 592 analysis. CG-C and MR-M acqui ed he unding. CP-G, CS-C, FM-P, MR-M and CG-C 593 e iewed and edi ed he pape . All au ho s p o ided eedback. 594 Re e ences 595 Ame ico da Sil a, T., B aunge , M.L., Ne is Cou inho, M.A., Rios do Ama al, L., 596 Rod igues, V. & Riul, A. (2019) 3D-P in ed G aphene Elec odes Applied in an 597 Impedime ic Elec onic Tongue o Soil Analysis. Chemosenso s, 7, 4, 50. 598 h p://dx.doi.o g/10.3390/chemosenso s7040050 599 Aouadi, B., Zinia Zaukuu, J.L., Vi ális, F., Bodo , Z., Fehé , O., Gillay, Z., Baza G. & 600 Ko acs, Z. (2020) His o ical E olu ion and Food Con ol Achie emen s o Nea In a ed 601 Spec oscopy, Elec onic Nose, and Elec onic Tongue—C i ical O e iew, Senso s, 602 20(19), A icle 5479. h p://dx.doi.o g/10.3390/s20195479 603 Çakı oğlu, B., Demi ci, Y.C., Gökgöz, E. & Özaca M. (2019) A Pho oelec ochemical 604 Glucose and Lac ose Biosenso Consis ing o Gold Nanopa icles, MnO2 and g-C3N4 605 deco a ed TiO2. Senso s and Ac ua o s B: Chemical, 282. 606 h p://dx.doi.o g/10.1016/j.snb.2018.11.064 607 Chiang, H., Wang, Y., Zhang, Q., & Le on, K. (2019). Op imiza ion o he 608 Elec odeposi ion o Gold Nanopa icles o he Applica ion o Highly Sensi i e, Label- 609 F ee Biosenso . Biosenso s, 9(2), 50. h ps://doi.o g/10.3390/bios9020050 610 Ciosek, P. & W oblewski, W. (2015) Po en iome ic And Hyb id Elec onic Tongues Fo 611 Biop ocess Moni o ing-An O e iew. Analy ical Me hods 7-9, 3958-3966 612 h p://dx.doi.o g/10.1039/c5ay00445d. 613 Ciosek, P. (2016) Milk and Dai y P oduc s Analysis by Means o an Elec onic Tongue. 614 In Elec onic Noses and Tongues in Food Science. Ed. Rod iguez-Mendez. Academic 615 P ess. (Chap e 212016), 209-223. h p://dx.doi.o g/10.1016/B978-0-12-800243- 616 8.00021-4 617 Collie , W.A., Bai d, D.B., Pa k-Ng, Z.A., Mo e, N. & Ha , A.L. (2003) Disc imina ion 618 among milks and cul u ed dai y p oduc s using sc een-p in ed elec ochemical a ays and 619 an elec onic nose, Senso s and Ac ua o s B: Chemical 92, 232–239. 620 h ps://doi.o g/10.1016/S0925-4005(03)00271-5 621 Co es, C. & Vapnik, V. (1995) Suppo - ec o ne wo ks. Machine Lea n 20, 273–297; 622 h p://dx.doi.o g/10.1007/BF00994018 623 Dias, L.A., Pe es, A.M., Veloso, A.C.A., Reis, F.S., Vilas-Boas, M. & Machado 624 A.A.S.C. (2009) An elec onic ongue as e e alua ion: iden i ica ion o goa milk 625 adul e a ion wi h bo ine milk. Senso s and Ac ua o s B: Chemical 136, 209–217. 626 h p://dx.doi.o g/10.1016/J.SNB.2008.09.025 627 Fox, P.F. & McSweeney, P.L.H. (1998) Dai y Chemis y and Biochemis y, Sp inge 628 Science & Business Media. 629 Jou nal P e-p oo 19 F iedman, J.H. (2002) S ochas ic g adien boos ing, Compu a ional S a is ics & Da a 630 Analysis, 30, 4, 367-378. h p://dx.doi.o g/10.1016/S0167-9473(01)00065-2 631 Ga cia-Cabezón, C., Gobbi, G., Dias, L.G., Sal o-Comino, C., Ga cía-He nandez, C., 632 Rod iguez-Mendez, M.L. & Ma in-Ped osa, F. (2020) Analysis o Phenolic Con en in 633 G ape Seeds and Skin by means o a Bio- Elec onic Tongue Using Elec ochemical 634 Modi ied Biosenso s wi h Ph halocyanines, NiO Nanopa icles and Ty osinase. Senso s, 635 20, 15, A icle 4176. h p://dx.doi.o g/10.3390%2Fs20154176 636 Ga cia-He nandez, C., Ga cia-Cabezon, C., Ma in-Ped osa, F., de Saja, J.A. & 637 Rod iguez-Mendez, M.L. (2019) Analysis o Mus s and Wines wi h a bio-elec onic 638 ongue based on Ty osinase and Glucose oxidase using a Polypy ole/gold nanopa icle 639 composi e as elec on media o . Food Chemis y, 289, 751-756. 640 h p://dx.doi.o g/10.1016/j. oodchem.2019.03.107 641 Ghasemi-Vamankhas i, M., Rod íguez-Méndez, M.L., Moh asebi, S.S., Ape ei, C., 642 Lozano, J., Raza i, S.H., Ahmadi, H. & de Saja, J.A. (2012) Moni o ing he aging o 643 bee s using a bioelec onic ongue, Food Con ol, 25, 216-224; 644 h p://dx.doi.o g/10.1016/j. oodcon .2011.10.020 645 Ha, D., Sun, Q.Y., Su, K., Wan, H., Haibo, L., Xu, N., Sun, F., Zhuang, L., Hu, N. & 646 Wang, P., (2015) Recen achie emen s in elec onic ongue and bioelec onic ongue as 647 as e senso s. Senso s and Ac ua o s B: Chemical, 207, 1136-1146; 648 h p://dx.doi.o g/10.1016/j.snb.2014.09.077 649 Haiss, W., Nguyen, T.T.K., A eya d, J. & Fe nig, D.G. (2007) De e mina ion o Size and 650 Concen a ion o Gold Nanopa icles om UV−Vis Spec a, Analy ical Chemis y 651 (2007), 4215-4221; h p://dx.doi.o g/10.1021/ac0702084 652 H uška , M., Majo , N., K pan, M. & Vahčić, N. (2010) Simul aneous de e mina ion o 653 e men ed milk a oma compounds by a po en iome ic senso a ay. Talan a. 82, 1292– 654 1297. h p://dx.doi.o g/10.1016/j. alan a.2010.06.048. 655 In e na ional O ganiza ion Fo S anda diza ion (2021) ISO/TC 34/SC 5 - Milk and Milk 656 P oduc s. h ps://www.iso.o g/commi ee/47878.h ml. 657 Jakha , S. & Pundi , C.S. (2018) P epa a ion, cha ac e iza ion and applica ion o u ease 658 nanopa icles o cons uc ion o an imp o ed po en iome ic u ea biosenso . 659 Biosenso s and Bioelec onics. 100, 242-250. 660 h p://dx.doi.o g/10.1016/j.bios.2017.09.005 661 Jia, W., Su, L. & Lei Y. (2011) P nano lowe /polyaniline composi e nano ibe s based 662 u ea biosenso . Biosenso s and Bioelec onics. 30, 1, 158-164. 663 h p://dx.doi.o g/10.1016/j.bios.2011.09.006 664 Juzhong & T., Jie, X. (2020). Applica ions o elec onic nose (e-nose) and elec onic 665 ongue (e- ongue) in ood quali y- ela ed p ope ies de e mina ion: A e iew, A i icial 666 In elligence in Ag icul u e, 4, 104-115. h p://dx.doi.o g/10.1016/j.aiia.2020.06.003 667 Kanyong, P., K ampa, F.D., Aniweh, Y. & Awanda e, G.A. (2019) Enzyme-based 668 ampe ome ic galac ose biosenso s: A e iew. Mic ochimica ac a, 184, 10, 3663-3671; 669 h p://dx.doi.o g/10.1007/s00604-017-2465-z 670 Kazenwadel, F., Wagne , H., Rapp, B.E., F anz eb, M. (2015) Op imiza ion O Enzyme 671 Immobiliza ion On Magne ic Mic opa icles Using 1-E hyl-3-(3- 672 Jou nal P e-p oo 20 Dime hylaminop opyl) Ca bodiimide (EDC) As A C osslinking Agen . Analy ical 673 Me hods. 7, 24, 10291-10298, h p://dx.doi.o g/10.1039/c5ay02670a 674 Kim, T.H., Song, H.S., Jin, H.J., Lee, S.H., Namgung, S., Kim, U.K., Pa k, T.H. & 675 Hong, S. (2011) Bioelec- onic supe - as e de ice based on as e ecep o -ca bon 676 nano ube hyb ids uc u es. Lab on a Chip, 11, 2262–2267. 677 h p://dx.doi.o g/10.1039/c0lc00648c 678 Kimling, J., Maie , M., Oken e, B., Ko aidis, V., Ballo , H. & Plech, A. (2006) Tu ke ich 679 Me hod o Gold Nanopa icle Syn hesis Re isi ed. J. Phys. Chem. B. 110, 15700-15707. 680 h p://dx.doi.o g/10.1021/jp061667w 681 Kuhn, M. (2008) Building P edic i e Models in R Using he ca e Package. Jou nal o 682 S a is ical So wa e, 28, 5, 1–26, h p://dx.doi.o g/10.18637/jss. 028.i05 683 Lee, J., Lee, I., Nam, J., Hwang, D.S, Yeon, K.M. & Kim, J. (2017) Immobiliza ion And 684 S abiliza ion O Acylase On Ca boxyla ed Polyaniline Nano ibe s Fo Highly E ec i e 685 An i ouling Applica ion Via Quo um Quenching, ACS Applied Ma e ials & In e aces, 686 9,18, 15424-15432; h p://dx.doi.o g/10.1021/acsami.7b01528 687 Li, L., Yu, Y., Yang, J., Yang, R., Dong, G. & Jin, T. (2015) Vol amme ic elec onic 688 ongue o he quali a i e analysis o milk adul e a ed wi h u ea combined wi h mul i- 689 way da a analysis. In e na ional Jou nal o Elec ochemis y. 10, 5970–5980. 690 h p://dx.doi.o g/10.1016/j.j oodeng.2018.09.022 691 Li, X., Li, S., Liu, Q., & Chen, Z. (2019) Elec onic-Tongue Colo ime ic-Senso A ay 692 o Disc imina ion and Quan i a ion o Me al Ions Based on Gold-Nanopa icle 693 Agg ega ion. Anal. Chem., 91, 9, 6315–6320. 694 h p://dx.doi.o g/10.1021/acs.analchem.9b01139 695 Mangan, D., McClea y, B.V., Culle on, H., Co naggio, C., I o y, R., McKie, V.A., 696 Delaney, E. & Ka gelis, T. (2018) A no el enzyma ic me hod o he measu emen o 697 lac ose in lac ose- ee p oduc s. Jou nal o he Science o Food and Ag icul u e. 99, 947- 698 956; h p://dx.doi.o g/10.1002/js a.9317 699 Me can e, L.A., Scagion, V.P., Pa ina o, A., San elice, R.C., Ma oso, L.H.C. & Co ea, 700 D.S. (2015) Elec onic ongue based on nanos uc u ed hyb id ilms o gold nanopa icles 701 and ph halocyanines o milk analysis, Jou nal o Nanoma e ials, 16, A icle 402; 702 h ps://doi.o g/10.1155/2015/890637 703 Miglio ini, F.L., San elice, R.C., Me can e, L.A., And e, R.S., Ma oso, L.H.C. & Co ea, 704 D.S. (2018) U ea impedime ic biosensing using elec ospun nano ibe s modi ied wi h 705 zinc oxide nanopa icles. Applied Su ace Science. 443, 18-23. 706 h p://dx.doi.o g/10.1016/j.apsusc.2018.02.168 707 Nguyen, B.H., Nguyen, B.T., Van Vu, H., Van Nguyen, C., Nguyen, D.T., Nguyen, L.T, 708 Thi Vu, T. & T an, L.D. (2016) De elopmen o label- ee elec ochemical lac ose 709 biosenso based on g aphene/poly(1,5-diaminonaph halene) ilm. Cu en Applied 710 Physics Volume 16, 2, 135-140. h p://dx.doi.o g/10.1016/j.cap.2015.11.004 711 Nguyen, H.H. & Kim, M. (2017) An o e iew o echniques in enzyme immobiliza ion. 712 Appl. Sci. Con e g. Technol. 26, 157-163. 713 h p://dx.doi.o g/10.5757/ASCT.2017.26.6.157 714 Ohlsson, J.A., Johansson, M., Hansson, H., Ab ahamson, A., Bybe g, L., Smedman, A., 715 Lindma k-Månsson, H. & Lundh, Å. (2017) Lac ose, glucose and galac ose con en in 716 Jou nal P e-p oo 21 milk, e men ed milk and lac ose- ee milk p oduc s, In e na ional Dai y Jou nal, 73, 151- 717 154. h p://dx.doi.o g/10.1016/j.idai yj.2017.06.004 718 Ohlsson, J.A., Johansson, M., Hansson, H., Ab ahamson, A., Bybe g, L., Smedman, A., 719 Lindma k-Månsson, H. & Lundh, Å. (2017) Lac ose, glucose and galac ose con en in 720 milk, e men ed milk and lac ose- ee milk p oduc s, In e na ional Dai y Jou nal. 73, 721 151-154. h p://dx.doi.o g/10.1016/j.idai yj.2017.06.004 722 Pascual, L., G as, M., Vidal-B o óns, D., Alcañiz, M., Ma ínez-Máñez, R. & Ros-Lis, 723 J. V. (2018) A ol amme ic e- ongue ool o he emula ion o he senso ial analysis and 724 he disc imina ion o ege al milks. Senso s and Ac ua o s B: Chemical, 270, 231-238. 725 h p://dx.doi.o g/10.1016/j.snb.2018.04.151 726 Pé ez-González, C., Sal o-Comino, C., Ma in-Ped osa, F., Dias, L., Rod iguez-Pe ez, 727 M.A., Ga cia-Cabezón, C. & Rod iguez-Mendez, M. L. (2021) Analysis O Milk Using 728 A Po able Po en iome ic Elec onic Tongue Based On Fi e Polyme ic Memb ane 729 Senso s, F on ie s In Chemis y 9. h p://dx.doi.o g/10.3389/ chem.2021.706460. 730 P ie o, N., Oli e i, P., Lea di, R., Gay, M., Ape ei, C., Rod iguez-Méndez, M.L. & de 731 Saja, J.A. (2013) Applica ion o a GA-PLS s a egy o a iable educ ion o elec onic 732 ongue signals. Senso s and Ac ua o s B: Chemical, Volume 183, 52- 57. 733 h p://dx.doi.o g/10.1016/j.snb.2013.03.114. 734 Rahman, M.M., Shiddiky, M.J.A., Rahman, M.A. & Shim, Y. (2009) A lac a e biosenso 735 based on lac a e dehyd ogenase/nic o inamide adenine dinucleo ide (oxidized o m) 736 immobilized on a conduc ing polyme /mul iwall ca bon nano ube composi e ilm. 737 Analy ical Biochemis y. 384, 1, 159-165. h p://dx.doi.o g/10.1016/j.ab.2008.09.030 738 Rod iguez-Mendez, M.L., De Saja, J.A. & González-An ón, R. (2016) Elec onic Noses 739 and Tongues in Wine Indus y. F on ie s in bioenginee ing and bio echnology, 4, 81. 740 h p://dx.doi.o g/10.3389/ bioe.2016.00081 741 Sal o-Comino, C., Ga cía-He nández, C., Ga cía-Cabezón, C. & Rod íguez-Méndez, 742 M.L. (2018) Disc imina ion o Milks wi h a Mul isenso Sys em Based on Laye -by- 743 Laye Films, Senso s, 18, 8, A icle 2716. h ps://doi.o g/10.3390/s18082716 744 Scagiona, V.P., Me can e, L.A., Sakamo o, K.Y., Oli ei a, J.E., Fonseca, F.J., Ma oso, 745 L.H.C., Fe ei a, M.D. & Co ea, D.S. (2016) An elec onic ongue based on conduc ing 746 elec ospun nano ibe s o de ec ing e acycline in milk samples, RSC Ad ., 6. 747 h p://dx.doi.o g/10.1039/C6RA21326J 748 Skladal, P. (2020) Sma bioelec onic ongues o ood and d inks con ol. T ac-T ends 749 In Analy ical Chemis y, 127, A icle 115887. 750 h p://dx.doi.o g/10.1016/j. ac.2020.115887 751 Sob ino-G ego io, L., Ba alle , R., So o, J. & Esc iche, I. (2018) Moni o ing honey 752 adul e a ion wi h suga sy ups using an au oma ic pulse ol amme ic elec onic ongue. 753 Food Con ol, 91, 254-260. h ps://doi.o g/10.1016/j. oodcon .2018.04.003. 754 Song, H.S., Kwon, O.S., Lee, S.H., Pa k, S.J., Kim, U.K., Jang, J. & Pa k, T.H. (2012) 755 Human as e ecep o - unc ionalized ield e ec ansis o as a human-like 756 nanobioelec onic ongue. Nano Le e s, 13, 172–178. 757 h p://dx.doi.o g/10.1021/nl3038147. 758 Tazi, I., T iyana, K., Siswan a, D., Veloso, A.C.A., Pe es, A.M. & Dias, L.G. (2018) 759 Dai y p oduc s disc imina ion acco ding o he milk ype using an elec ochemical 760 Jou nal P e-p oo 22 mul isenso de ice coupled wi h chemome ic ools, Jou nal o Food Measu emen and 761 Cha ac e iza ion 12, 2385–2393. h p://dx.doi.o g/10.1007/s11694-018-9855-8. 762 Teodo o, K.B.R., Shimizu, F.M., Scagion, V.P. & Co ea, D. S. (2019) Te na y 763 nanocomposi es based on cellulose nanowhiske s, sil e nanopa icles and elec ospun 764 nano ibe s: Use in an elec onic ongue o hea y me al de ec ion. Senso s and Ac ua o s 765 B: Chemical, 290, 387-395; h p://dx.doi.o g/10.1016/j.snb.2019.03.125 766 Theodo e, B.T. & Robin, C.G. (2006) Robus classi ica ion and eg ession using 767 suppo ec o machines, Eu opean Jou nal o Ope a ional Resea ch, 173, 3, 893-909. 768 h p://dx.doi.o g/10.1016/j.ejo .2005.07.024. 769 Tian, S., Chen, M., Pan, W. & Deng, S. (2013) A compa ison s udy o h ee nonlinea 770 mul i a ia e da a analysis me hods in sma ongue: Ke nel PCA LLE and Sammon 771 mapping. J. Chem. Pha m. Res, 5, 575-783. ISSN : 0975-7384. 772 Vaghela, C., Kulka ni, M., Ha am, S., Aiye , R., & Ka e, M. (2018). A no el inhibi ion 773 based biosenso using u ease nanoconjuga e en apped biocomposi e memb ane o 774 po en iome ic glyphosa e de ec ion. In e na ional Jou nal O Biological 775 Mac omolecules, 108, 32-40. h ps://doi.o g/10.1016/j.ijbiomac.2017.11.136 776 Valen e, N.I.P., Rudni skaya, A., Oli ei a, J.A.B.P., Gaspa , E.M.M. & Gomes, M.T.S.R. (2018) Qua z 777 c ys als Cheeses Made om Raw and Pas eu ized Cow’s Milk Analysed by an Elec onic Nose and an 778 Elec onic Tongue, Senso s 18,8, A icle 2415. h ps://doi.o g/10.3390/s18082415 779 Wang, Q. & del Valle, W. (2021) De e mina ion o Chemical Oxygen Demand (COD) 780 Using Nanopa icle-Modi ied Vol amme ic Senso s and Elec onic Tongue P inciples. 781 Chemosenso s, 9, 3, 46. h p://dx.doi.o g/10.3390/chemosenso s9030046 782 Wasilewski, T., Kamysz, W. & Gebicki, J. (2020) Bioelec onic ongue: Cu en s a us 783 and pe spec i es, Biosenso s & Bioelec onics, 150, 11, A icle 1923; 784 h p://dx.doi.o g/10.1016/j.bios.2019.111923 785 Wei, Z., Wang, J. & Jin, W. (2013) E alua ion o a ie ies o se yogu s and hei 786 physical p ope ies using a ol amme ic elec onic ongue based on a ious po en ial 787 wa e o ms Senso s and Ac ua o s B: Chemical. 177, 684-694. 788 h p://dx.doi.o g/10.1016/j.snb.2012.11.056 789 Winquis , F., K an z-Rulcke , C., Wide, P. & Lunds om, I. (1998) Moni o ing o esh 790 milk, by an elec onic ongue on he basis o ol amme y. Measu emen Science and 791 Technology, 9, 1937–1946Z. h ps://doi.o g/10.1088/0957-0233%2F9%2F12%2F002 792 Xiaoyan, W.P.Z. (2003) Model Selec ion o SVM wi h RBF Ke nel and i s Applica ion. 793 Compu e Enginee ing and Applica ions, 24, 021. h p://dx.doi.o g/10.1007/11875581_6 794 Yang, Z., Miao, N., Zhang, X., Li, Q., Wang, Z., Li, C., Sun, X. & Lan,Y. (2021) 795 Employmen o an elec onic ongue combined wi h deep lea ning and ans e lea ning 796 o disc imina ing he s o age ime o Pu-e h ea, Food Con ol. 121, 107608. 797 h ps://doi.o g/10.1016/j. oodcon .2020.107608. 798 Yhan, S.M., Rosa io, D., Sil a, L.R.G., San os, F.D., San os, L.P., Janegi z, B.C., 799 Filguei as, P.R., Romão, W., Fe ei a, R.Q. & Con e-Junio , C.A. (2021) Po able 800 elec onic ongue based on sc een-p in ed elec odes coupled wi h chemome ics o apid 801 di e en ia ion o B azilian lage bee . Food Con ol, 127, A icle 108163. 802 h ps://doi.o g/10.1016/j. oodcon .2021.108163 803 Yu, Y., Zhao, H., Yang, R., Dong, G., Li, L., Yang, J., Jin, T., Zhang, W., Liu & Y. (2015) 804 Jou nal P e-p oo 23 Pu e milk b ands classi ica ion by means o a ol amme ic elec onic ongue and 805 mul i a ia e analysis. In e na ional Jou nal o Elec ochemis y, 10, 4381–4392. 806 Zhou, Z. H. (2012) Ensemble me hods: ounda ions and algo i hms. CRC p ess. 807 Zucca, P. & Sanjus , E. (2014) Ino ganic Ma e ials As Suppo s Fo Co alen Enzyme 808 Immobiliza ion: Me hods And Mechanisms. Molecules. 19, 9, 14139-14194. 809 h p://dx.doi.o g/10.3390/molecules190914139 810 811 812 Biog aphies 813 814 Cla a Pe ez-Gonzalez ob ained he Ms in Nanoscience in 2019 (U. Valladolid. Spain). 815 She is cu en ly wo king on he PhD Thesis which is dedica ed o he de elopmen o 816 elec ochemical senso s o he analysis o oods. She is au ho o 5 scien i ic pape s. 817 818 Co al Sal o-Comino ob ained he Ms in Analy ical Chemis y in 2015 (U. Complu ense. 819 Mad id. Spain). She is cu en ly wo king on he PhD Thesis which is dedica ed o he 820 de elopmen o elec ochemical senso s o he analysis o oods. She is au ho o 13 821 scien i ic pape s. 822 823 Fe nando Ma in-Ped osa is ull p o esso a he Uni e si y o Valladolid and Head o he 824 Depa men o Ma e ials Science. His esea ch is dedica ed o elec ochemis y s udies o 825 di e en solid ma e ials. He is au ho o mo e han 80 pape s. 826 827 C is ina Ga cia Cabezón, is assis an p o esso a he Enginee s school o he Uni e si y 828 o Valladolid. She is an expe in elec ochemis y and impedance spec oscopy. She is 829 au ho o coau ho o mo e han 50 pape s in he ield. 830 831 Ma ia Luz Rod iguez-Mendez is Full p o esso o Ino ganic Chemis y a he Enginee s 832 School o he Uni e si y o Valladolid and Head o he g oup o senso s UVASens. She 833 is leading se e al unded P ojec s de o ed o he de elopmen o a ays o ol amme ic 834 nanos uc u ed senso s and biosenso s o he cha ac e iza ion o oods. She is au ho o 835 co-au ho o o e 165 publica ions (H index 44), se en books and h ee pa en s in he 836 ield. 837 838 839 840 841 842 843 844 845 846 847 Jou nal P e-p oo