![\"\"](\"https:\/\/www.hashtag.al\/wp-content\/uploads\/2024\/04\/1198900295.jpg.0-ezgif.com-webp-to-jpg-converter.jpg\")
<\/p>\n<\/p>\n
D\u00ebshmia m\u00eb e habitshme se inteligjenca artificiale mund t\u00eb siguroj\u00eb zbulime t\u00eb thella shkencore erdh\u00ebn me zbulimin e nj\u00eb programi t\u00eb quajtur \u2018AlphaFold\u2019 nga \u2018Google DeepMind\u2019. <\/span><\/p>\n N\u00eb vitin 2016, studiuesit n\u00eb kompani kishin sh\u00ebnuar nj\u00eb sukses t\u00eb madh me \u2018AlphaGo\u2019, nj\u00eb sistem AI (Artificial Intelligence) i cili, pasi i m\u00ebsoi vetes rregullat e Go, vijoi t\u00eb mundte lojtar\u00ebt njer\u00ebz m\u00eb t\u00eb vler\u00ebsuar t\u00eb loj\u00ebs, ndonj\u00ebher\u00eb duke p\u00ebrdorur taktika q\u00eb askush nuk i kishte parashikuar ndonj\u00ebher\u00eb.<\/span><\/p>\n Kjo e nxiti kompanin\u00eb t\u00eb nd\u00ebrtonte nj\u00eb sistem q\u00eb do p\u00ebrpunonte nj\u00eb s\u00ebr\u00eb rregullash shum\u00eb m\u00eb t\u00eb nd\u00ebrlikuara: ato p\u00ebrmes t\u00eb cilave sekuenca e aminoacideve q\u00eb p\u00ebrcakton nj\u00eb protein\u00eb t\u00eb ve\u00e7ant\u00eb \u00e7on n\u00eb form\u00ebn n\u00eb t\u00eb cil\u00ebn paloset sekuenca kur ajo protein\u00eb prodhohet. AlphaFold i gjeti ato rregulla dhe i zbatoi me sukses t\u00eb jasht\u00ebzakonsh\u00ebm.<\/span><\/p>\n Arritja ishte e jasht\u00ebzakonshme dhe e dobishme. E mahnitshme sepse shum\u00eb njer\u00ebz t\u00eb zgjuar ishin p\u00ebrpjekur o\u00ebr shum\u00eb koh\u00eb p\u00ebr t\u00eb krijuar modele kompjuterike t\u00eb proceseve q\u00eb palosin zinxhir\u00ebt e aminoacideve n\u00eb proteina p\u00ebr dekada. \u2018AlphaFold\u2019 b\u00ebri p\u00ebrpjekjet m\u00eb t\u00eb mira m\u00ebnyr\u00eb po aq t\u00eb plot\u00eb sa sistemi Go q\u00eb e frym\u00ebzoi i mposht lojtar\u00ebt njer\u00ebz. <\/span><\/p>\n E dobishme sepse forma e nj\u00eb proteine \u200b\u200bka nj\u00eb r\u00ebnd\u00ebsi t\u00eb madhe praktike: ajo p\u00ebrcakton se \u00e7far\u00eb b\u00ebn proteina dhe \u00e7far\u00eb mund t’i b\u00ebjn\u00eb molekulat e tjera. T\u00eb gjitha proceset themelore t\u00eb jet\u00ebs varen nga ajo q\u00eb b\u00ebjn\u00eb proteinat specifike. Gjetja e molekulave q\u00eb u b\u00ebjn\u00eb gj\u00ebra t\u00eb d\u00ebshirueshme proteinave (nganj\u00ebher\u00eb bllokojn\u00eb veprimin e tyre, ndonj\u00ebher\u00eb duke e inkurajuar at\u00eb) \u00ebsht\u00eb q\u00ebllimi i shumic\u00ebs d\u00ebrrmuese t\u00eb programeve t\u00eb zhvillimit t\u00eb barnave n\u00eb bot\u00eb.<\/span><\/p>\n P\u00ebr shkak t\u00eb r\u00ebnd\u00ebsis\u00eb s\u00eb struktur\u00ebs tredimensionale t\u00eb proteinave, ekziston nj\u00eb n\u00ebndisiplin\u00eb e t\u00ebr\u00eb e p\u00ebrkushtuar kryesisht ndaj saj: biologjia strukturore. Ajo p\u00ebrdor t\u00eb gjitha llojet e teknologjis\u00eb p\u00ebr t\u00eb par\u00eb proteinat p\u00ebrmes teknikave t\u00eb rezonanc\u00ebs b\u00ebrthamore-magnetike ose duke i kristalizuar ato (gj\u00eb q\u00eb mund t\u00eb jet\u00eb shum\u00eb e v\u00ebshtir\u00eb) dhe duke i shp\u00ebrthyer ato me rreze x.<\/span><\/p>\n P\u00ebrpara \u2018AlphaFold\u2019-it, mbi gjysm\u00eb shekulli i biologjis\u00eb strukturore kishte prodhuar disa qindra mij\u00eb struktura proteinike t\u00eb besueshme n\u00ebp\u00ebrmjet k\u00ebtyre mjeteve. \u2018AlphaFold\u2019 dhe rival\u00ebt e tij (m\u00eb s\u00eb shumti nj\u00eb program i b\u00ebr\u00eb nga Meta) tani kan\u00eb ofruar parashikime t\u00eb holl\u00ebsishme t\u00eb formave.<\/span><\/p>\n Si m\u00ebnyr\u00eb q\u00eb i la shkenc\u00ebtar\u00ebt t\u00eb hutuar, \u00ebsht\u00eb nj\u00eb veprim i v\u00ebshtir\u00eb p\u00ebr t’u ndjekur. Por n\u00ebse produktet e \u2018AlphaFold\u2019 kan\u00eb mahnitur bot\u00ebn, bazat se si e b\u00ebri jan\u00eb mjaft tipike p\u00ebr ato lloj gj\u00ebrash q\u00eb m\u00ebsimi i thell\u00eb dhe ai gjenerues mund t\u00eb ofrojn\u00eb p\u00ebr biologjin\u00eb.<\/span><\/p>\n I st\u00ebrvitur me mbi dy lloje t\u00eb ndryshme t\u00eb dh\u00ebnash (sekuencat aminoacide dhe p\u00ebrshkrimet tredimensionale t\u00eb formave n\u00eb t\u00eb cilat palosen) AlphaFold gjeti modele q\u00eb e lejuan t\u00eb p\u00ebrdorte llojin e par\u00eb t\u00eb t\u00eb dh\u00ebnave p\u00ebr t\u00eb parashikuar t\u00eb dytin. Parashikimet nuk jan\u00eb t\u00eb gjitha perfekte. <\/span><\/p>\n Chris Gibson, shefi i \u2018Recursion Pharmaceuticals\u2019, nj\u00eb startup intensiv i zbulimit t\u00eb barnave me baz\u00eb n\u00eb Utah, SHBA, thot\u00eb se kompania e tij i trajton rezultatet e \u2018AlphaFold\u2019 si hipoteza p\u00ebr t’u testuar dhe v\u00ebrtetuar eksperimentalisht. Dr. Gibson thot\u00eb se modeli po p\u00ebrmir\u00ebsohet shpejt.<\/span><\/p>\n \u00cbndrrat e kristalta<\/strong><\/span><\/p>\n Kjo \u00ebsht\u00eb ajo q\u00eb nj\u00eb gam\u00eb e t\u00ebr\u00eb l\u00ebvizjesh po b\u00ebjn\u00eb tani n\u00eb bot\u00ebn e biomjek\u00ebsis\u00eb dhe, ve\u00e7an\u00ebrisht, n\u00eb k\u00ebrkimin e barnave: duke b\u00ebr\u00eb sugjerime p\u00ebr m\u00ebnyr\u00ebn se si \u00ebsht\u00eb bota q\u00eb shkenc\u00ebtar\u00ebt mund ose nuk do arrinin vet\u00eb.<\/span><\/p>\n T\u00eb st\u00ebrvitur p\u00ebr t\u00eb gjetur modele q\u00eb shtrihen n\u00ebp\u00ebr trupa t\u00eb m\u00ebdhenj t\u00eb dh\u00ebnash t\u00eb ndryshme, sistemet AI mund t\u00eb zbulojn\u00eb marr\u00ebdh\u00ebnie brenda atyre t\u00eb dh\u00ebnave q\u00eb kan\u00eb implikime p\u00ebr biologjin\u00eb dhe s\u00ebmundjet njer\u00ebzore. T\u00eb ngarkuar me t\u00eb dh\u00ebna t\u00eb reja, ata mund t\u00eb p\u00ebrdorin ato modele t\u00eb implikimit p\u00ebr t\u00eb prodhuar hipoteza t\u00eb reja t\u00eb cilat m\u00eb pas mund t\u00eb testohen.<\/span><\/p>\n Aft\u00ebsia e AI p\u00ebr t\u00eb gjeneruar ide t\u00eb reja u ofron p\u00ebrdoruesve njohuri q\u00eb mund t\u00eb ndihmojn\u00eb n\u00eb identifikimin e objektivave t\u00eb barnave dhe p\u00ebr t\u00eb parashikuar sjelljen e p\u00ebrb\u00ebrjeve t\u00eb reja, ndonj\u00ebher\u00eb t\u00eb pap\u00ebrfytyruara m\u00eb par\u00eb, q\u00eb mund t\u00eb veprojn\u00eb si barna. Gjithashtu p\u00ebrdoret p\u00ebr t\u00eb gjetur aplikime t\u00eb reja p\u00ebr barnat e vjetra, p\u00ebr t\u00eb parashikuar efektet an\u00ebsore t\u00eb barnave t\u00eb reja dhe p\u00ebr t\u00eb gjetur m\u00ebnyra p\u00ebr t’u treguar pacient\u00ebve q\u00eb nj\u00eb bar mund t\u00eb ndihmoj\u00eb nga d\u00ebme t\u00eb ndryshme. <\/span><\/p>\n Ambicie t\u00eb tilla llogarit\u00ebse nuk jan\u00eb t\u00eb reja. Informatika n\u00eb shkall\u00eb t\u00eb gjer\u00eb, m\u00ebsimi i makinerive dhe dizajni i ila\u00e7eve po bashkoheshin q\u00eb n\u00eb vitet 2000, thot\u00eb Vijay Pande, i cili ishte studiues n\u00eb Universitetin e Stanfordit n\u00eb at\u00eb koh\u00eb. Kjo ishte pjes\u00ebrisht nj\u00eb p\u00ebrgjigje ndaj gjetjeve t\u00eb reja n\u00eb l\u00ebmin e biologjis\u00eb: tani ka m\u00eb shum\u00eb se nj\u00eb milion dokumente k\u00ebrkimore biomjek\u00ebsore t\u00eb botuara \u00e7do vit.<\/span><\/p>\n Nj\u00eb nga m\u00ebnyrat e hershme n\u00eb t\u00eb cilat ai u pa se mund t\u00eb ndihmonte ishte p\u00ebrmes “grafik\u00ebve t\u00eb njohurive”, t\u00eb cilat lejuan q\u00eb i gjith\u00eb ai informacion t\u00eb lexohej nga makinerit\u00eb dhe t\u00eb shqyrtohej p\u00ebr t\u00eb kuptuar, t\u00eb themi, cilat proteina n\u00eb gjak mund t\u00eb p\u00ebrdoreshin si bioshenjues q\u00eb zbulonin pranin\u00eb os\u00eb ashp\u00ebrsin\u00eb e nj\u00eb s\u00ebmundjeje. N\u00eb vitin 2020, \u2018Benevolentai\u2019, me qend\u00ebr n\u00eb Lond\u00ebr, p\u00ebrdori k\u00ebt\u00eb metod\u00eb p\u00ebr t\u00eb par\u00eb potencialin q\u00eb baricitinib, i shitur nga Eli Lilly si nj\u00eb trajtim p\u00ebr artritin reumatoid, kishte p\u00ebr trajtimin e Covid-19.<\/span><\/p>\n K\u00ebt\u00eb janar, hulumtimi i publikuar n\u00eb Science p\u00ebrshkruan se si algoritmet e nj\u00eb lloji t\u00eb ndrysh\u00ebm kishin p\u00ebrshpejtuar p\u00ebrpjekjet p\u00ebr t\u00eb gjetur bioshenjues t\u00eb Covid-it n\u00eb gjak. Qasjet statistikore p\u00ebr zbulimin e bioshenjuesve t\u00eb till\u00eb mund t\u00eb jen\u00eb sfiduese duke pasur parasysh kompleksitetin e t\u00eb dh\u00ebnave. <\/span><\/p>\n AI ofron nj\u00eb m\u00ebnyr\u00eb p\u00ebr t\u00eb zbutur k\u00ebt\u00eb gj\u00eb dhe p\u00ebr t\u00eb \u00e7uar p\u00ebrpara procesin e zbulimit n\u00eb s\u00ebmundje si t\u00eb reja, si Covid-i i gjat\u00eb, ashtu edhe s\u00ebmundjet m\u00eb t\u00eb v\u00ebshtira p\u00ebr t’u diagnostikuar, si fazat e hershme t\u00eb Alzheimerit.<\/span><\/p>\n Koha e duhur<\/strong><\/span><\/p>\n Por pavar\u00ebsisht k\u00ebtij progresi n\u00eb t\u00eb kaluar\u00ebn, Dr Pande, tani n\u00eb \u2018Andreessen Horo\u00ebitz\u2019, nj\u00eb firm\u00eb e kapitalit sip\u00ebrmarr\u00ebs q\u00eb \u00ebsht\u00eb me pesh\u00eb n\u00eb AI, mendon se p\u00ebrparimet m\u00eb t\u00eb fundit sh\u00ebnojn\u00eb nj\u00eb ndryshim t\u00eb madh. K\u00ebrkimet biomjek\u00ebsore, ve\u00e7an\u00ebrisht n\u00eb bioteknologji dhe farmaceutik\u00eb, po e rrisnin n\u00eb m\u00ebnyr\u00eb t\u00eb q\u00ebndrueshme var\u00ebsin\u00eb e saj n\u00eb automatizimin dhe inxhinierin\u00eb p\u00ebrpara se modelet e reja t\u00eb dilnin n\u00eb vetvete; tani kjo ka ndodhur, t\u00eb dy duket se p\u00ebrforcojn\u00eb nj\u00ebri-tjetrin. <\/span><\/p>\n Modelet e reja t\u00eb themelimit nuk ofrojn\u00eb vet\u00ebm nj\u00eb m\u00ebnyr\u00eb p\u00ebr t’u p\u00ebrballur me sasi t\u00eb m\u00ebdha t\u00eb t\u00eb dh\u00ebnave; i k\u00ebrkojn\u00eb ato. T\u00eb dh\u00ebnat e besueshme q\u00eb laborator\u00ebt shum\u00eb t\u00eb automatizuar mund t\u00eb prodhojn\u00eb me boll\u00ebk jan\u00eb vet\u00ebm nj\u00eb lloj gj\u00ebje p\u00ebr modelet e themeleve t\u00eb trajnimit. Dhe studiuesit biomjek\u00ebsor\u00eb kan\u00eb nevoj\u00eb p\u00ebr t\u00eb gjith\u00eb ndihm\u00ebn q\u00eb mund t\u00eb marrin p\u00ebr t\u00eb kuptuar rrjedhat e t\u00eb dh\u00ebnave q\u00eb tani jan\u00eb n\u00eb gjendje t\u00eb gjenerojn\u00eb.<\/span><\/p>\n Biologjia mund t\u00eb konsiderohet si “nj\u00eb sistem p\u00ebrpunimi i informacionit, megjith\u00ebse jasht\u00ebzakonisht kompleks”.<\/span><\/p>\n Duke gjetur modele q\u00eb njer\u00ebzit nuk kishin menduar t\u00eb k\u00ebrkonin, ose nuk kishin shpres\u00eb p\u00ebr t’i gjetur pa ndihm\u00eb, ai u ofron studiuesve m\u00ebnyra t\u00eb reja p\u00ebr t\u00eb eksploruar dhe kuptuar misteret e jet\u00ebs. Disa flasin p\u00ebr zot\u00ebrimin e “gjuh\u00ebs s\u00eb biologjis\u00eb”, t\u00eb m\u00ebsuarit p\u00ebr t\u00eb kuptuar at\u00eb q\u00eb evolucioni ka krijuar drejtp\u00ebrdrejt nga t\u00eb dh\u00ebnat n\u00eb t\u00eb nj\u00ebjt\u00ebn m\u00ebnyr\u00eb q\u00eb, t\u00eb trajnuar n\u00eb shum\u00eb gjuh\u00eb reale, ata mund t\u00eb gjenerojn\u00eb rrjedhsh\u00ebm fjali kuptimplote t\u00eb pashqiptuara kurr\u00eb m\u00eb par\u00eb.<\/span><\/p>\n Demis Hassabis, shefi i \u2018DeepMind\u2019, thekson se vet\u00eb biologjia mund t\u00eb konsiderohet si “nj\u00eb sistem p\u00ebrpunimi i informacionit, megjith\u00ebse nj\u00eb sistem jasht\u00ebzakonisht kompleks dhe dinamik”. N\u00eb nj\u00eb postim n\u00eb Medium, Serafim Batzoglou, shefi i t\u00eb dh\u00ebnave n\u00eb \u2018Seer Bio\u2019, nj\u00eb kompani e Silicon Valley q\u00eb \u00ebsht\u00eb e specializuar n\u00eb k\u00ebqyrjen se si sillen proteinat, parashikon shfaqjen e modeleve t\u00eb hapura q\u00eb do integrojn\u00eb t\u00eb dh\u00ebnat q\u00eb shtrihen q\u00eb nga sekuencat e gjenomit n\u00eb historit\u00eb mjek\u00ebsore. K\u00ebto, argumenton ai, do p\u00ebrshpejtojn\u00eb shum\u00eb inovacionin dhe do avancojn\u00eb mjek\u00ebsin\u00eb precize.<\/span><\/p>\n Ashtu si shum\u00eb nga entuziast\u00ebt e AI, Dr. Pande flet p\u00ebr nj\u00eb “revolucion industrial\u2026duke ndryshuar gjith\u00e7ka”. Por t\u00eb kuptuarit e tij p\u00ebr koh\u00ebn e kaluar se \u00e7on at\u00eb t\u00eb kujdeset se arritjet q\u00eb justifikojn\u00eb at\u00eb ndryshim afatgjat\u00eb t\u00eb entuziazmit nuk do vijn\u00eb brenda nat\u00ebs: “Ne jemi n\u00eb nj\u00eb periudh\u00eb kalimtare ku njer\u00ebzit mund t\u00eb shohin ndryshimin, por ka ende pun\u00eb p\u00ebr t\u00eb b\u00ebr\u00eb.”<\/span><\/p>\n T\u00eb gjitha t\u00eb dh\u00ebnat nga kudo n\u00eb t\u00eb nj\u00ebjt\u00ebn koh\u00eb<\/strong><\/span><\/p>\n Shum\u00eb firma farmaceutike kan\u00eb b\u00ebr\u00eb investime t\u00eb konsiderueshme n\u00eb zhvillimin e modeleve n\u00eb vitet e fundit. Krahas k\u00ebsaj ka qen\u00eb nj\u00eb rritje n\u00eb startup-et t\u00eb p\u00ebrqendruara tek AI, si \u2018Recursion\u2019, \u2018Genesis Therapeutics\u2019, me baz\u00eb n\u00eb Silicon Valley, \u2018Insilico\u2019, me baz\u00eb n\u00eb Hong Kong dhe Nju Jork dhe \u2018Relay Therapeutics\u2019, n\u00eb Kembrixh, Massachusetts. <\/span><\/p>\n Daphne Koller, shefe e \u2018Insitro\u2019, nj\u00eb kompani bioteknologjike n\u00eb San Francisko, thot\u00eb se nj\u00eb shenj\u00eb e koh\u00ebs \u00ebsht\u00eb se ajo nuk ka m\u00eb nevoj\u00eb t\u00eb shpjegoj\u00eb modele t\u00eb m\u00ebdha gjuh\u00ebsore dhe t\u00eb m\u00ebsuarit e vet\u00ebmbik\u00ebqyrur. Dhe Nvidia – e cila b\u00ebn nj\u00ebsit\u00eb e p\u00ebrpunimit grafik q\u00eb jan\u00eb thelb\u00ebsore p\u00ebr fuqizimin e modeleve t\u00eb themeleve – ka treguar nj\u00eb interes t\u00eb madh.<\/span><\/p>\n Vitin e kaluar, ajo ka investuar ose ka b\u00ebr\u00eb marr\u00ebveshje partneriteti me t\u00eb pakt\u00ebn gjasht\u00eb firma t\u00eb ndryshme bioteknike t\u00eb p\u00ebrqendruara te AI, p\u00ebrfshir\u00eb \u2018Schrodinger\u2019, nj\u00eb tjet\u00ebr firm\u00eb me baz\u00eb n\u00eb Nju Jork, \u2018Genesis\u2019, \u2018Recursion\u2019 dhe \u2018Genentech\u2019, nj\u00eb deg\u00eb e pavarur e \u2018Roche\u2019, nj\u00eb kompani e madhe farmaceutike zvicerane.<\/span><\/p>\n Modelet e zbulimit t\u00eb barnave me t\u00eb cilat punojn\u00eb shum\u00eb kompani mund t\u00eb m\u00ebsojn\u00eb nga nj\u00eb shum\u00ebllojshm\u00ebri e gjer\u00eb t\u00eb dh\u00ebnash biologjike, duke p\u00ebrfshir\u00eb sekuencat e gjeneve, fotografit\u00eb e qelizave dhe indeve, strukturat e proteinave p\u00ebrkat\u00ebse, bioshenjuesit n\u00eb gjak, proteinat q\u00eb prodhohen n\u00eb qeliza specifike dhe t\u00eb dh\u00ebnat klinike p\u00ebr rrjedh\u00ebn e s\u00ebmundjes dhe efektin e trajtimeve te pacient\u00ebt. Pasi t\u00eb trajnohet, kanali mund t\u00eb akordohet mir\u00eb me t\u00eb dh\u00ebna t\u00eb etiketuara p\u00ebr t\u00eb p\u00ebrmir\u00ebsuar aft\u00ebsit\u00eb e tyre.<\/span><\/p>\n P\u00ebrdorimi i t\u00eb dh\u00ebnave t\u00eb pacientit \u00ebsht\u00eb ve\u00e7an\u00ebrisht interesant. P\u00ebr arsye mjaft t\u00eb dukshme, shpesh nuk \u00ebsht\u00eb e mundur t\u00eb zbulohet funksionimi i sakt\u00eb i nj\u00eb s\u00ebmundjeje te njer\u00ebzit p\u00ebrmes eksperimentit. Pra, zhvillimi i barnave zakonisht mb\u00ebshtetet shum\u00eb n\u00eb modelet e kafsh\u00ebve, edhe pse ato mund t\u00eb jen\u00eb t\u00eb rreme. Ato q\u00eb jan\u00eb trajnuar dhe p\u00ebrshtatur m\u00eb mir\u00eb me biologjin\u00eb njer\u00ebzore mund t\u00eb ndihmojn\u00eb n\u00eb shmangien e disa rrug\u00ebve qorre q\u00eb pengojn\u00eb zhvillimin e barit.<\/span><\/p>\n \u2018Insitro\u2019, p\u00ebr shembull, trajnon modelet e saj mbi patologjit\u00eb, sekuencat e gjeneve, t\u00eb dh\u00ebnat e MRI dhe proteinat e gjakut. Nj\u00eb nga modelet e tij \u00ebsht\u00eb n\u00eb gjendje t\u00eb lidh\u00eb ndryshimet n\u00eb pamjen e qelizave n\u00ebn mikroskop me mutacionet themelore n\u00eb gjenom \u00a0me rezultatet klinike n\u00eb s\u00ebmundje t\u00eb ndryshme. Kompania shpreson t\u00eb p\u00ebrdor\u00eb k\u00ebto dhe teknika t\u00eb ngjashme p\u00ebr t\u00eb gjetur m\u00ebnyra p\u00ebr t\u00eb identifikuar n\u00ebngrupet e pacient\u00ebve me kancer q\u00eb do ken\u00eb sukses ve\u00e7an\u00ebrisht n\u00eb kurse specifike t\u00eb trajtimit.<\/span><\/p>\n \u2018Recursion\u2019 thot\u00eb se mund t\u00eb p\u00ebrdor\u00eb robotik\u00eb laboratorik\u00eb p\u00ebr t\u00eb kryer 2.2 milion\u00eb eksperimente \u00e7do jav\u00eb.<\/span><\/p>\n Ndonj\u00ebher\u00eb t\u00eb zbulosh se cilit aspekt t\u00eb t\u00eb dh\u00ebnave po i p\u00ebrgjigjet nj\u00eb AI \u00ebsht\u00eb i dobish\u00ebm n\u00eb vetvete. N\u00eb vitin 2019, O\u00ebkin, nj\u00eb \u201cAI biotech\u201d me baz\u00eb n\u00eb Paris, publikoi holl\u00ebsi t\u00eb nj\u00eb rrjeti t\u00eb thell\u00eb nervor t\u00eb trajnuar p\u00ebr t\u00eb parashikuar mbijetes\u00ebn n\u00eb pacient\u00ebt me mesothelioma malinje, nj\u00eb kancer i indit q\u00eb rrethon mushk\u00ebrit\u00eb, n\u00eb baz\u00eb t\u00eb mostrave t\u00eb indeve t\u00eb k\u00ebqyrura. <\/span><\/p>\n AI zbuloi se qelizat m\u00eb t\u00eb p\u00ebrshtatshme p\u00ebr parashikimet nuk ishin vet\u00eb qelizat e kancerit, por qelizat jo kanceroze aty pran\u00eb. Ekipi O\u00ebkin solli t\u00eb dh\u00ebna shtes\u00eb qelizore dhe molekulare n\u00eb foto dhe zbuloi nj\u00eb objektiv t\u00eb ri t\u00eb barit. N\u00eb gusht t\u00eb vitit t\u00eb kaluar, nj\u00eb ekip shkenc\u00ebtar\u00ebsh nga Universiteti i Indian\u00ebs, Bloomington trajnuan nj\u00eb model mbi t\u00eb dh\u00ebnat se si qelizat e kancerit reagojn\u00eb ndaj barnave (p\u00ebrfshir\u00eb informacionin gjenetik) dhe strukturave kimike t\u00eb barnave, duke e lejuar at\u00eb t\u00eb parashikoj\u00eb se sa efektiv do ishte nj\u00eb ila\u00e7 n\u00eb trajtimin e nj\u00eb kanceri specifik.<\/span><\/p>\n Shum\u00eb prej kompanive q\u00eb p\u00ebrdorin AI kan\u00eb nevoj\u00eb p\u00ebr v\u00ebllime t\u00eb m\u00ebdha t\u00eb dh\u00ebnash me cil\u00ebsi t\u00eb lart\u00eb, saq\u00eb po i gjenerojn\u00eb ato vet\u00eb si pjes\u00eb e programeve t\u00eb tyre t\u00eb zhvillimit t\u00eb barnave n\u00eb vend q\u00eb t\u00eb presin q\u00eb ato t\u00eb publikohen diku tjet\u00ebr. Nj\u00eb variant i k\u00ebsaj teme vjen nga nj\u00eb nj\u00ebsi e re e shkencave kompjuterike n\u00eb \u2018Genentech\u2019, e cila p\u00ebrdor nj\u00eb qasje “laboratori” p\u00ebr t\u00eb trajnuar AI-n\u00eb e saj. <\/span><\/p>\n Parashikimet e sistemit testohen n\u00eb nj\u00eb shkall\u00eb t\u00eb gjer\u00eb me an\u00eb t\u00eb eksperimenteve t\u00eb kryera me sisteme t\u00eb automatizuara laboratorike. Rezultatet e k\u00ebtyre eksperimenteve p\u00ebrdoren m\u00eb pas p\u00ebr t\u00eb ritrajnuar ai dhe p\u00ebr t\u00eb rritur sakt\u00ebsin\u00eb e tij. \u2018Recursion\u2019, i cili p\u00ebrdor nj\u00eb strategji t\u00eb ngjashme, thot\u00eb se mund t\u00eb p\u00ebrdor\u00eb robotik\u00eb t\u00eb automatizuar laboratorik\u00eb p\u00ebr t\u00eb kryer 2.2 milion\u00eb eksperimente \u00e7do jav\u00eb.<\/span><\/p>\n \u00c7\u00ebshtja \u00ebsht\u00eb ndryshimi<\/strong><\/span><\/p>\n Nd\u00ebrsa firmat farmaceutike b\u00ebhen gjithnj\u00eb e m\u00eb t\u00eb uritura p\u00ebr t\u00eb dh\u00ebna, shqet\u00ebsimet p\u00ebr privat\u00ebsin\u00eb e t\u00eb dh\u00ebnave t\u00eb pacient\u00ebve po b\u00ebhen m\u00eb t\u00eb theksuara. Nj\u00eb m\u00ebnyr\u00eb p\u00ebr t’u marr\u00eb me problemin, e p\u00ebrdorur nga O\u00ebkin nd\u00ebr t\u00eb tjera, \u00ebsht\u00eb “t\u00eb m\u00ebsuarit e federuar”, n\u00eb t\u00eb cilin t\u00eb dh\u00ebnat e trajnimit q\u00eb i nevojiten p\u00ebr t\u00eb nd\u00ebrtuar nj\u00eb atlas t\u00eb llojeve t\u00eb qelizave kancerogjene nuk largohen kurr\u00eb nga spitali ku ruhen mostrat e indeve t\u00eb k\u00ebrkuara: \u00e7far\u00eb t\u00eb dh\u00ebnash mund t\u00eb ofroj\u00eb n\u00eb drejtim t\u00eb trajnimit \u00ebsht\u00eb hequr. Vet\u00eb t\u00eb dh\u00ebnat mbeten.<\/span><\/p>\n Implikimet e AI shkojn\u00eb p\u00ebrtej t\u00eb kuptuarit t\u00eb s\u00ebmundjes dhe m\u00eb tej n\u00eb zbulimin se si t\u00eb nd\u00ebrhyhet. Modelet gjeneruese t\u00eb AI, t\u00eb tilla si \u2018Proteinsgm\u2019 nga Universiteti i Torontos, tani jan\u00eb mjete t\u00eb fuqishme n\u00eb dizajnimin e proteinave, sepse ato nuk jan\u00eb thjesht n\u00eb gjendje t\u00eb p\u00ebrshkruajn\u00eb proteinat ekzistuese, por edhe t\u00eb dizajnojn\u00eb t\u00eb reja – me karakteristikat e d\u00ebshiruara – q\u00eb aktualisht nuk ekzistojn\u00eb n\u00eb natyr\u00eb, por q\u00eb jan\u00eb m\u00ebnyra t\u00eb mundshme p\u00ebr t\u00eb mish\u00ebruar nj\u00eb funksion t\u00eb d\u00ebshiruar. Sisteme t\u00eb tjera u lejojn\u00eb kimist\u00ebve t\u00eb dizajnojn\u00eb molekula t\u00eb vogla q\u00eb mund t\u00eb jen\u00eb t\u00eb dobishme si ila\u00e7e pasi ato nd\u00ebrveprojn\u00eb me nj\u00eb objektiv n\u00eb nj\u00eb m\u00ebnyr\u00eb t\u00eb d\u00ebshiruar.<\/span><\/p>\n N\u00eb \u00e7do faz\u00eb, hipotezat e AI duhet t\u00eb kontrollohen. Megjithat\u00eb, nj\u00eb qasje e till\u00eb duket se p\u00ebrshpejton zbulimin. Nj\u00eb analiz\u00eb e fundit e barnave nga firmat “ai-intensive” e kryer nga BCG, nj\u00eb grup k\u00ebshillues, zbuloi se nga tet\u00eb barna p\u00ebr t\u00eb cilat kishte informacion, pes\u00eb kishin arritur n\u00eb provat klinike n\u00eb m\u00eb pak se koha e zakonshme. <\/span><\/p>\n Punime t\u00eb tjera sugjerojn\u00eb se ai mund t\u00eb sjell\u00eb kursime n\u00eb koh\u00eb dhe kosto prej 25% deri n\u00eb 50% n\u00eb faz\u00ebn paraklinike t\u00eb zhvillimit t\u00eb barnave, e cila mund t\u00eb zgjas\u00eb kat\u00ebr deri n\u00eb shtat\u00eb vjet. Duke pasur parasysh koston n\u00eb koh\u00eb dhe para t\u00eb t\u00eb gjith\u00eb procesit, q\u00eb mund t\u00eb jet\u00eb disa miliarda dollar\u00eb p\u00ebr nj\u00eb ila\u00e7 t\u00eb vet\u00ebm, p\u00ebrmir\u00ebsimet mund t\u00eb shnd\u00ebrrojn\u00eb produktivitetin e industris\u00eb. Por do duhet koh\u00eb p\u00ebr ta ditur me siguri.<\/span><\/p>\n \u2018Insilico Medicine\u2019 \u00ebsht\u00eb nj\u00eb nga kompanit\u00eb q\u00eb shpreson q\u00eb kjo t\u00eb ndryshoj\u00eb. Ai p\u00ebrdor nj\u00eb s\u00ebr\u00eb modelesh n\u00eb procesin e tij t\u00eb zhvillimit t\u00eb barnave. Nj\u00ebra identifikon proteinat q\u00eb mund t\u00eb synohen t\u00eb ndikojn\u00eb n\u00eb nj\u00eb s\u00ebmundje. Nj\u00eb tjet\u00ebr mund t\u00eb projektoj\u00eb p\u00ebrzierje t\u00eb reja t\u00eb mundshme t\u00eb ila\u00e7it. Duke p\u00ebrdorur k\u00ebt\u00eb qasje, ai identifikoi nj\u00eb kandidat p\u00ebr ila\u00e7 q\u00eb mund t\u00eb jet\u00eb i dobish\u00ebm kund\u00ebr fibroz\u00ebs pulmonare n\u00eb m\u00eb pak se 18 muaj dhe me nj\u00eb kosto prej 3 milion\u00eb dollar\u00ebsh. Ila\u00e7i filloi s\u00eb fundmi provat e Faz\u00ebs 2.<\/span><\/p>\n Shum\u00eb firma farmaceutike n\u00eb Kin\u00eb po b\u00ebjn\u00eb marr\u00ebveshje me kompani t\u00eb drejtuara nga AI si \u2018Insilico\u2019 me shpres\u00ebn p\u00ebr t\u00eb par\u00eb m\u00eb shum\u00eb t\u00eb nj\u00ebjt\u00ebn gj\u00eb. Disa shpresojn\u00eb se marr\u00ebveshje t\u00eb tilla mund t\u00eb jen\u00eb n\u00eb gjendje t\u00eb nxisin bizneset e zhvillimit relativisht t\u00eb ngadalt\u00eb t\u00eb Kin\u00ebs p\u00ebr zhvillimin e drog\u00ebs. <\/span><\/p>\n Organizatat k\u00ebrkimore t\u00eb Kin\u00ebs tashm\u00eb po ndjejn\u00eb p\u00ebrfitimet e interesit t\u00eb nxitur nga AI p\u00ebr molekula t\u00eb reja nga e gjith\u00eb bota. Investimi n\u00eb zbulimin e barit me ndihm\u00ebn e AI n\u00eb Kin\u00eb ishte m\u00eb shum\u00eb se 1.26 miliard\u00eb dollar\u00eb n\u00eb 2021.<\/span><\/p>\n Bota ka par\u00eb nj\u00eb s\u00ebr\u00eb barnash dhe trajtimesh t\u00eb reja t\u00eb reja n\u00eb dekad\u00ebn e fundit: barna q\u00eb synojn\u00eb glp-1 q\u00eb po shnd\u00ebrrojn\u00eb trajtimin e diabetit dhe obezitetit; terapit\u00eb car-t q\u00eb aktivizojn\u00eb sistemin imunitar kund\u00ebr kancerit; aplikimet e para klinike t\u00eb redaktimit t\u00eb gjenomit. <\/span><\/p>\n Por periudha e gjat\u00eb e zhvillimit t\u00eb barnave, nga dallimi i proceseve biologjike q\u00eb kan\u00eb r\u00ebnd\u00ebsi n\u00eb identifikimin e objektivave t\u00eb barnave, deri te zhvillimi i molekulave candidate e deri te vendosja e tyre p\u00ebrmes testeve paraklinike dhe m\u00eb pas provave klinike, mbetet p\u00ebrgjith\u00ebsisht pun\u00eb e ngadalt\u00eb dhe zhg\u00ebnjyese. <\/span><\/p>\n P\u00ebraf\u00ebrsisht 86% e t\u00eb gjith\u00eb kandidat\u00ebve p\u00ebr barna t\u00eb zhvilluara nd\u00ebrmjet viteve 2000 dhe 2015 nuk arrit\u00ebn t\u00eb p\u00ebrmbushin pikat kryesore p\u00ebrfundimtare n\u00eb provat klinike. Vitet e ardhshme do tregojn\u00eb p\u00ebrfundimisht n\u00ebse AI \u00ebsht\u00eb n\u00eb gjendje ta ndryshoj\u00eb k\u00ebt\u00eb pamje. <\/span><\/p>\n N\u00ebse ofron vet\u00ebm p\u00ebrmir\u00ebsime n\u00eb rritje, mund t\u00eb jet\u00eb ende nj\u00eb ndihm\u00eb e madhe. N\u00ebse lejon q\u00eb biologjia t\u00eb deshifrohet n\u00eb nj\u00eb m\u00ebnyr\u00eb krejt\u00ebsisht t\u00eb re, si\u00e7 sugjerojn\u00eb m\u00eb p\u00ebrforcuesit, mund ta b\u00ebj\u00eb t\u00eb gjith\u00eb procesin shum\u00eb m\u00eb t\u00eb suksessh\u00ebm dhe m\u00eb efikas. <\/span><\/p>\n Analist\u00ebt n\u00eb BCG shohin shenja t\u00eb nj\u00eb vale t\u00eb medikamenteve t\u00eb reja q\u00eb po afrohet me shpejt\u00ebsi. Dr. Pande paralajm\u00ebron se rregullator\u00ebt e drog\u00ebs do duhet t\u00eb p\u00ebrmir\u00ebsojn\u00eb loj\u00ebn e tyre p\u00ebr t\u00eb p\u00ebrballuar sfid\u00ebn. Do ishte nj\u00eb risi e mir\u00eb p\u00ebr bot\u00ebn.<\/span><\/p>\n